Start with a diagnostic. Sit a full past paper for every subject under exam conditions, mark it honestly, and write down which chapters lost you the most points. Do not skip subjects you feel confident about — confidence is the least reliable signal you have.

Divide the remaining days into three blocks. The first sixty days are for repair: the specific chapters your diagnostic flagged, in the order that closes the biggest point gaps first. Not the order they appear in the textbook. Not the order they will appear on the exam. Biggest gap first.

The next twenty days are for pattern practice: full sections under time, then targeted drills on the question shapes that ate your minutes. This is when you build your 错题本 in earnest — one page per class of mistake, not one per question.

The final two weeks are taper. No new material. Old mock papers re-marked. Sleep hits nine hours a night. The goal is to arrive on June 7 as the same person who has been quietly getting better for three months, not as someone in the middle of a sprint.

Move the base unit to fifty minutes on, ten off, and keep the phone in another room during the fifty. The break is not for scrolling. Stand up, drink water, look at something twenty meters away.

After three cycles, take a longer break — thirty minutes, ideally outside. This is the part everyone skips and it is the part that keeps you honest for six hours instead of two.

One useful modification for math and physics: if you are mid-problem when the timer goes off, finish the problem. The rule exists to serve the work, not the other way around.

The alternative is embarrassingly simple: close the book and write, from memory, everything you think the chapter said. Do it on blank paper. Then open the book and mark, in a different color, everything you missed or got wrong.

For math and physics, replace 'write' with 'do': pick one hard problem you have not seen and solve it without notes. If you can't, that is the diagnostic you needed.

Do this after every study session, not once a week. The pain of realizing you already forgot Monday's material is the mechanism.

If you find yourself reaching for a highlighter, stop. Highlighting is note-taking that pretends to be studying.

Any spaced-repetition system will work: Anki, a paper Leitner box with five compartments, or index cards you shuffle through on the bus. The tool is not the point. The schedule is.

Add no more than twenty new items per day. If you add fifty because you feel behind, tomorrow's reviews will crush you and you will quit the system by the end of the week.

For classical Chinese, do not just memorize the character. Memorize a short line from a canonical text that uses it in the sense you care about. Meaning without context evaporates within a week.

Pick one notebook per subject and one system per subject. Cornell-style works well for humanities (cue column on the left, notes on the right, one-line summary at the bottom). For math and physics, a single wide column with worked solutions is enough.

Never copy from the textbook into your notes. If it is already written down somewhere you own, do not write it again — reference the page number.

Reserve the last five pages of each notebook for your 错题本 entries in that subject. This keeps mistakes close to the material they came from.

The useful version is organized by the class of mistake, not the individual question. 'Forgot to check the domain' is a class. 'Confused ∫ tan x dx' is a class. Under each class, you paste in one or two representative wrong questions and the fix that made them click.

Review it not by re-reading, but by covering the answer and re-doing the question. Two weeks after adding an entry, if you get it right cold twice in a row, retire the entry.

The point of the 错题本 is to become smaller over time. If it is only growing, it is a diary, not a study tool.

Before spending another week on a topic you keep failing, compute the ratio: how many points does this topic represent on the exam, and how many hours would getting good at it take you? If the answer is 'four points and thirty hours', bank the four and move on.

This is not the same as giving up. It is deciding that your time is finite and there is a physics chapter next door where the same thirty hours might buy you fifteen points.

The rule works in both directions: sometimes a small topic you have written off is worth revisiting because you finally have the base to understand it.

Once a week, sit a full past paper with a real clock, no phone, no snacks, no bathroom breaks except at the halfway mark. Mark it the next day, coldly.

The first few times you do this you will be shocked at how much slower you are than in casual practice. This is the useful shock. It is far better to feel it in April than in June.

Track two numbers over the term: total score, and time spent on the section that gave you the most trouble. Both should trend in the right direction. If one is improving and the other is not, adjust.

If every mock produces a huge emotional swing, something is off. Either the mocks are wildly different in difficulty, or you are treating each one as a referendum on your entire future.

The individual score does not matter. The trend across five mocks matters. Track it. If it is flat, look at what you are doing between mocks and change one thing.

One useful trick: after each mock, write down the one specific thing you will do differently on the next one. Just one. Fixing everything at once fixes nothing.

Do sit two full papers per subject over these two weeks under exam conditions. Mark them. Re-do only the questions you got wrong, one time each.

Re-read your 错题本 twice this fortnight. That is the highest-leverage reading you can do.

Sleep goes up, not down. Nine hours a night, non-negotiable. This alone is worth several points on paper 1.

Three days before the exam, stop doing full papers. Skim your notes, review the 错题本 one last time, pack your bag, walk to the exam center once so you know the route.

Start with a one-sentence thesis in the opening paragraph. State the position you are taking. Do not begin with a rhetorical question — every fourth essay begins that way and none of them earn full marks.

Each body paragraph should have one clear claim, one strong example (historical, literary, or contemporary — vary these across the three), and one sentence tying the example back to your thesis. Do not use the same kind of example three times in a row.

The closing paragraph is not a summary. It is where you elevate the thesis into a wider point about society, character, or knowledge. One sentence of aphorism, quoted or your own. Then stop.

Handwriting matters. It matters more than students want to believe. Not calligraphy — legibility, consistent size, straight lines. If a grader has to strain to read you, you are already down a point.

Skim the questions first — thirty seconds. Now read the passage with a pen, underlining every point that seems to answer one of them. When you get to a question, jump back to your underlines.

For subjective questions ('作者的态度是……'), answer in the vocabulary the passage uses. Graders are looking for specific phrases from the text. Paraphrasing loses points even when the paraphrase is correct.

Never leave a subjective question blank. Even a weak answer that echoes one keyword from the passage earns a point. Blank earns zero.

Judgment sentences: '……者，……也' and its variants. Recognize the frame and you can translate the noun-is-noun claim without hesitation.

Passive constructions: '为……所……' and '见……于……'. These trip students who default to reading them actively. Slow down every time you see 为 or 见 as the second character in a clause.

Object inversion in interrogative and negative sentences: '何……之有?' means 'what … is there?'. Recognize the shape before you attempt a word-for-word translation.

Omitted subjects are everywhere. When the meaning does not make sense, supply the missing subject before assuming you have misidentified a word.

Practice with the same twenty passages until you can read them cold. It is worth more than reading two hundred once.

Memorize, cold: the derivatives of xⁿ, eˣ, ln x, sin x, cos x, and tan x. The chain rule, product rule, and quotient rule. Say them out loud once a day for a week and they stop being a bottleneck.

Applications almost always take one of three shapes: find the tangent line at a point, find intervals of monotonicity, or find extrema. Practice one of each per day.

The classic trap in monotonicity questions is forgetting to check the endpoints of the interval and the points where the derivative is undefined. Build the habit of listing every candidate before deciding which is the maximum.

For extrema on a closed interval, always compute f at both endpoints as well as at every interior critical point. Skipping this step is the most common lost mark on this topic.

The first move is always to write the curve in standard form: x²/a² + y²/b² = 1 for an ellipse, x²/a² - y²/b² = 1 for a hyperbola. Identify a, b, and c (with c² = a² ± b²). Half the work of most problems is done here.

For chord-length problems, substitute the line into the curve, use Vieta's formulas for x₁+x₂ and x₁x₂, then apply the chord-length formula |AB| = √(1+k²) · |x₁-x₂|. This template solves an enormous fraction of the questions on this topic.

For focus-related questions, the definition (sum or difference of distances to the two foci) is often faster than coordinate bashing. When the question mentions a focus, always try the definition first.

Draw the curve. Every time. A sketch of an ellipse takes ten seconds and stops you from writing a hyperbola formula for an ellipse question.

Two questions to ask first: is this with or without replacement, and is order relevant? These two answers determine whether you are counting combinations, permutations, or something else entirely.

For conditional probability, write out P(A|B) = P(A∩B) / P(B) explicitly before plugging numbers. Half the mistakes on this topic come from computing P(A∩B) and calling it P(A|B).

Expected value questions on the gaokao almost always reduce to E(X) = Σ xᵢ · pᵢ. Build the distribution table before writing the formula.

For the normal-distribution question at the end of the section, memorize the three intervals (μ±σ ≈ 68.3%, μ±2σ ≈ 95.4%, μ±3σ ≈ 99.7%). The question tests whether you can read a table, not integrate.

For each passage, spend twenty seconds on the first and last sentence of every paragraph. That gives you the argument's skeleton. Now read the questions and only re-read the passage in the specific places you need.

For 'main idea' questions, the answer is almost always paraphrased from the first or last paragraph. If a candidate answer is a paraphrase of a middle-paragraph detail, it is a distractor.

For inference questions, be conservative: the right answer is only mildly beyond what the passage said. If a candidate answer is a huge leap, it is wrong even if it sounds insightful.

Vocabulary-in-context questions: substitute the candidate word into the original sentence and see if the sentence still makes sense in the passage's argument. Do not rely on your dictionary definition.

Read the whole passage first without looking at the candidates. Get the argument.

Now look at each gap. What comes before it? What comes after it? Underline the pronouns and connectives near each gap — 'this', 'however', 'in addition' — these are the seams the correct candidate has to fit.

Match candidates to gaps by these seams, not by topic. Two sentences might be about the same topic but only one will grammatically join what comes before and after.

Do the easiest gaps first. The last gap you fill in should be the hardest one, and by then you have only two candidates left to choose between.

Paragraph one: state your purpose in one sentence. 'I am writing to invite you to…', 'I am writing to apply for…'. Do not warm up. Do not open with a greeting question ('How are you?').

Paragraph two: give the specifics — dates, times, expectations, requests. Two or three sentences. Use linking words: 'firstly', 'in addition', 'therefore'.

Paragraph three: close politely. 'I would be grateful if you could confirm your attendance.' 'Looking forward to hearing from you.' Then 'Yours sincerely / Yours faithfully' — sign off with the correct one.

Memorize a bank of ten polite formulas. Rotate them across practice essays so no two of yours start the same way. Graders notice.

For an object on an inclined plane, resolve gravity into components along and perpendicular to the incline. Along: mg sin θ. Perpendicular: mg cos θ. From these two, most inclined-plane problems reduce to F = ma along the incline.

For pulley problems with two connected masses, the constraint is that the acceleration magnitudes are equal (assuming inextensible rope and frictionless pulley). Write one Newton's second law equation for each mass. Solve the system.

Friction on the incline: static friction points up the slope when the object is about to slide down, and vice versa. Kinetic friction opposes actual motion. Get the direction right before writing the coefficient.

Sanity check: does your acceleration point in the physically sensible direction? If a heavier mass on one side of a pulley 'accelerates upward' in your answer, you have a sign error.

Step one: identify the magnetic flux through the loop, Φ = BA cos θ. Determine which of B, A, or θ is changing.

Step two: apply Faraday's law to get the induced EMF, ε = -dΦ/dt. If a rod is sliding on rails at speed v through a field B, this simplifies to ε = BLv.

Step three: compute the induced current, I = ε/R. Get the sign right by Lenz's law: the induced current opposes the change in flux.

Step four: if the question asks about force, use F = BIL for the force on the current-carrying conductor. This force will oppose the motion that induced the current in the first place.

Energy questions on this topic almost always reduce to 'work done against the magnetic force equals electrical energy dissipated in the resistor'. Write that equation before computing anything.

Atomic radius: decreases across a period, increases down a group. Reason: more protons pull the same shell tighter across a period; a new shell adds distance going down a group.

Ionization energy: increases across a period, decreases down a group. Reason: harder to remove an electron from a tightly-held small atom; easier from a large, loosely-held outer shell.

Electronegativity: increases across a period, decreases down a group. Fluorine is the peak. Cesium is near the floor.

Metallic character: decreases across a period, increases down a group — the mirror of electronegativity.

Any exam question on trends reduces to picking the right two elements and applying one of these four rules. Do not try to memorize specific values.

Substitution: one atom or group replaces another (e.g., halogenation of alkanes with light: CH₄ + Cl₂ → CH₃Cl + HCl). Free-radical mechanism.

Addition: unsaturated bond opens up and two groups attach (e.g., alkene + H₂ → alkane; alkene + HBr → haloalkane). Markovnikov's rule applies to asymmetric addition.

Elimination: opposite of addition — two groups leave, forming an unsaturated bond (e.g., alcohol + concentrated H₂SO₄ heat → alkene + H₂O).

Esterification: carboxylic acid + alcohol ⇌ ester + water, catalyzed by concentrated H₂SO₄, reversible. The classic exam favorite.

Oxidation: alcohol → aldehyde → carboxylic acid, with CuO or acidified KMnO₄ as oxidant.

For each type, memorize one canonical example. When a question appears, ask: what functional group is present, and which of these five reactions can it undergo?

Inputs: photosynthesis takes CO₂ and H₂O; respiration takes glucose (C₆H₁₂O₆) and O₂.

Outputs: photosynthesis produces glucose and O₂; respiration produces CO₂ and H₂O. Yes, they are exact reverses of one another at the summary level.

Location: photosynthesis in chloroplasts (light reactions on thylakoid membranes; Calvin cycle in the stroma). Aerobic respiration in mitochondria (glycolysis actually in cytoplasm; Krebs in matrix; oxidative phosphorylation on inner membrane).

Energy currency: both use ATP and NADPH/NADH. Photosynthesis generates them from light; respiration generates ATP by consuming them.

Exam questions often ask about conditions where only one process runs (dark for a plant, or in an animal cell). Once you can hold the comparison in one image, these questions become one-step.

Monohybrid: Aa × Aa gives 1 AA : 2 Aa : 1 aa. Phenotypes: 3 dominant : 1 recessive. Memorize this.

Dihybrid: AaBb × AaBb gives 9 A_B_ : 3 A_bb : 3 aaB_ : 1 aabb. This is the famous 9:3:3:1 ratio and it comes from the two genes assorting independently.

Test cross: an unknown-genotype dominant crossed with a homozygous recessive. If any recessive offspring appear, the unknown is heterozygous.

Sex-linked traits (X-linked recessive) appear more often in males. If a trait skips a generation and appears in a grandson but not the mother, it is likely X-linked recessive.

Always write out the parental genotypes explicitly before drawing the Punnett square. The mistakes at this step are the ones that ruin the whole problem.

You cannot bring a phone into the exam room. Not switched off in your bag. Not at all. Detection is by walk-through metal detectors and hand-held wands. A watch is allowed only if it is a simple analog or digital timepiece — no smart features.

You will be given a national identity card and a 准考证 (admission ticket). Both are checked at the door. Losing either means a same-morning phone call to a parent to bring backups. Some centers keep spare printouts of admission tickets — do not rely on this.

The room is silent to a degree most students have never experienced. This is not a mood — it is enforced. No pencil tapping. No sighing. No talking, even in the last thirty seconds while you finish. Practice mock papers under conditions this strict.

You may not leave the room in the last thirty minutes of any session. Plan bathroom breaks accordingly. When the bell rings you must stop writing immediately — the invigilator's discretion for lateness is zero.

After each session you walk out and, ideally, do not discuss the paper with anyone. Regret ruins the next session more reliably than any exam question can.

The free-response section is graded by humans who scan sheet images. Write inside the printed boxes. Anything outside the box is invisible to the grader.

You may not use correction fluid or tape on the answer sheet. If you make a mistake, cross it out cleanly and rewrite. Do not scribble.

For the essay, count characters as you go — most provinces require at least 800 characters, and running short costs more than the missing content.

Leave the last five minutes of each subject strictly for transferring answers from the question booklet to the answer sheet if you have been working on scratch. This is when the most preventable errors happen.

Most provinces now use a 平行志愿 system: you rank up to about ten universities in order of preference, and the system slots you into the highest-ranked one that will admit you at your score. Ordering matters.

The classic strategy is 冲 (reach), 稳 (target), 保 (safety). Two or three reach schools whose historical cutoff is slightly above your score. Four or five target schools where you are safely admissible. Two or three safeties well below your score.

Look at the historical cutoff scores (投档线) for the past three years, not just last year. Cutoffs bounce. A one-year dip does not make a school a safety.

Read the enrollment plan (招生计划) for your major of interest. Some majors have only a handful of seats. A university you can get into does not guarantee the major you want in that university.

Do not choose a major only because its cutoff is high. High cutoff often means everyone else also thought it was prestigious, not that the program is best for you.

The 批次线 (batch cutoff) is set by your province and determines which tier of universities (一本 / 二本 / 专科) you can apply to at all. Below the batch cutoff, that whole tier is closed to you regardless of specific-school interest.

The 投档线 (school cutoff) is set retroactively for each university and major, based on the lowest score of admitted students that year. This is the one you actually need to beat to get into a specific place.

Historical 投档线 for the past three years is your best signal. Universities publish these. A gap between your score and the historical cutoff of about 10-20 points is a healthy target-school margin.

Regional differences are enormous. A score that is 一本 in one province is 二本 in another. Never compare your score to a national average.

In practice, the top-ranked national universities all admit as 一本, and the correlation between tier and reputation is strong at the top. But there are excellent programs at nominally 二本 schools, particularly in engineering and applied fields.

The new gaokao provinces (新高考) have been gradually merging 一本 and 二本 into a single batch called 本科批, replacing 'tier' with a smoother continuous admission process. Check which system your province uses.

The 双一流 (double first class) designation, introduced in 2017, marks a set of universities and disciplines the state has selected for elite funding. It is a more current signal of national investment than the old 985/211 lists.

For most students, the useful shorthand is: which universities have your target major, and where do their historical cutoffs fall? Tier is a starting filter, not the answer.

In '3+1+2', all students take Chinese, math, and English. They then choose one of physics or history as their primary elective (the '1'), and two more from among chemistry, biology, geography, and political science (the '2').

In '3+3', the same three core subjects plus three chosen freely from a wider pool. Shanghai and Zhejiang piloted this format.

The chosen subjects are scored by percentile-conversion (等级赋分) rather than raw points. This is meant to fairly compare, say, a chemistry student and a geography student — but it introduces a strategic layer.

Universities and majors declare subject requirements. A physics-track student can apply to most STEM majors; a history-track student is largely restricted to humanities and business. Choose your electives with your intended major in mind, in tenth grade.

By 2026, the majority of provinces are on some form of the new gaokao. Check your province's specific implementation — the details matter.

Henan sends more than a million candidates per year and receives a comparatively small share of top-university seats. The score required to enter, say, Peking University from Henan is meaningfully higher than the score required from Beijing.

This is not a secret and it is not accidental. Universities administered by a city or province tend to enroll more students from that region, and the biggest universities are concentrated in a handful of provinces.

The upshot is that your gaokao score is only meaningful in comparison to other students in your province. National ranking lists exist but are not what universities actually use.

One consequence: transferring your household registration to move for the gaokao is legally and practically difficult, and the state actively discourages 'high-migration' (高考移民).

Two 2B pencils, sharpened. A pencil sharpener with a shavings-catcher. A soft eraser.

Two black-ink ballpoint pens. Not blue. Not gel unless allowed by your province.

A transparent ruler, a set of drawing compasses, and a set square for math and physics sessions.

A water bottle with the label removed. Some centers do not permit any labeled containers.

A tissue pack and a small hand towel. The rooms in June can be warm.

Nothing with electronics: no watch that vibrates, no smart bracelet, no wireless earbuds in your pocket. Not even switched off.

Pack the bag the night before and check it once in the morning. Do not repack in the morning — that is when you forget things.

5.7 million people sat the first restored gaokao in the winter of 1977. Fewer than 5% were admitted to university. The candidate pool included people in their thirties who had waited a decade for another chance.

Through the 1980s and 1990s, the exam gradually standardized. National paper sets, provincial adaptations, the introduction of English as a core subject, and the 3+X reform of 1999 (which added a chosen subject to the three core subjects) all reshaped what students studied.

In 1999 the state also began the massive expansion of higher education. University enrollments grew several-fold in a decade, and the gaokao shifted from being a filter for a tiny elite to being a sorting mechanism for a large majority of high-school leavers.

The new gaokao reform, from 2014 onward, is the most recent major change: elective subjects, percentile scoring, and the gradual merger of the tier system. As of 2026, most provinces are on some form of the new system.

The cohort included former Red Guards, sent-down youth returning from the countryside, factory workers, and teenagers straight from high school. There was no age limit that first year.

The pass rate was brutally low, but for those who succeeded, the transformation was total. This cohort — 77级 — filled Chinese universities in a period of great intellectual openness and went on to disproportionately shape the country's academic, political, and business elites of the 1990s and 2000s.

The subsequent 78级 sat their exam in the summer of 1978, restoring the traditional summer schedule. Together, 77 and 78 are treated as a single historical cohort in Chinese cultural memory.

Films, memoirs, and novels about this period are numerous. It remains one of the most-referenced moments in the country's post-Mao history.

But the same studies also show the mobility engine grinding slower over time. Students from the top decile of household income are dramatically over-represented at elite universities relative to their share of the population.

Reasons include the cost and quality gap between urban and rural high schools, the tutoring industry (now legally curtailed but not eliminated), and the provincial quota system that favors students in provinces where the top universities are located.

Recent policy responses include targeted enrollment programs (国家专项计划, 高校专项计划, 地方专项计划) that reserve seats at elite universities for students from designated rural areas. These have helped at the margins but have not closed the gap.

The debate about whether the gaokao is a fair mobility engine, a lottery, or a reproduction mechanism for existing inequalities is one of the most active discussions in Chinese education policy.

Both exams are (or were) standardized, meritocratic in principle, and function as the main gate through which young people from ordinary families can enter the elite. Both breed intensive preparation cultures.

But the differences are large. The keju tested classical scholarship — memorization of the Confucian canon and the ability to write formal essays on canonical themes. The gaokao tests a broad academic curriculum including math, science, and a foreign language.

The keju was a job interview: passing at the top level (进士) qualified you for imperial office. The gaokao is a university admissions exam. There is no direct government appointment at the end of it.

The keju admitted a few thousand candidates a year, empire-wide. The gaokao admits millions. Their social function is comparable at a very high level of abstraction and diverges everywhere else.

The 1980s: idealistic and pedagogical. Prompts asked students to reflect on famous quotations, on the value of study, on the importance of contributing to society.

The 1990s: more open-ended, sometimes surprising. The 1999 national prompt — 'if memory could be transplanted' — became famous for its speculative, almost science-fictional cast.

The 2000s: broad thematic prompts, often accompanied by a short parable to interpret. The essay became a test of both reasoning and general reading.

The 2010s: contemporary and often data-driven. Prompts about the internet, about the environment, about the role of the individual in a technological society.

The 2020s: heavy on narratives of national development, quotations from leadership speeches, and prompts that expect students to connect personal choice to national trajectory.

Reading old prompts is a genuinely useful essay-preparation exercise — it trains you to see the shape of the exam's ideal answer before you start writing.

《高考1977》(2009) dramatizes the restoration of the gaokao and the sent-down-youth generation. Formal, warm-hued, historical.

《青春派》(2013) is a comedy about a class of graduating seniors preparing for the exam. Lighter than the topic usually gets.

《全城高考》(2013) follows several students and their teacher in the run-up to the exam. Melodramatic in the good way.

《学区房72小时》(2019) is not about the gaokao itself but about the school-district housing market that surrounds it — a useful sideways look.

The Chinese-Indian co-production 《嗝嗝老师》 (an adaptation of Hichki, 2018) touches on similar exam-culture themes and has found a large Chinese audience.

Watch one, not all. And not the week of the exam.

Aim for at least seven hours a night through most of the year and at least eight in the last month. If you cannot make this work with your current schedule, cut something else — not sleep.

The exception is not 'the night before an important test'. That is exactly the night sleep matters most. A well-rested student outperforms a slightly-better-prepared exhausted one on almost every measurable dimension of exam performance.

Naps of 20-30 minutes in the afternoon are a real study tool, not a luxury. They restore attention and consolidate what you learned in the morning.

Coffee helps for two hours and hurts for eight. Time it deliberately. Do not drink coffee after 2 PM in the last month if you value your sleep.

It helps to recognize what parents actually want. Under the surface of grade demands, most parents want their child to have more options than they had. The demand is instrumental — the underlying goal is not a specific score.

It also helps to name the pressure explicitly, once. A calm conversation early in the year — 'I know you're worried, I'm working as hard as I can, I need you to trust me' — often reduces the daily friction more than any tactic.

Try not to argue about grades in the hour after a mock exam result. The emotional charge on both sides is too high. Sleep, then talk the next day.

If the pressure crosses into something you cannot cope with — sleeplessness, panic attacks, an inability to work — reach out. A school counselor, a teacher, or a helpline is the right move, not a private struggle.

Eat the meals you always eat, at the times you always eat them. Familiar bowls of rice, familiar noodles, familiar vegetables. This is not the week to experiment.

Breakfast on exam day should include some protein (egg, milk, soy milk) and some slow carbohydrate (steamed bun, congee). Avoid heavy oil — it slows you down.

Do not drink so much liquid at breakfast that you need the bathroom mid-exam. Sip through the morning instead.

Chocolate is not a magic exam food. Bananas are not either. Eat what you have always eaten, in slightly smaller portions than usual.

Do not open the paper immediately. Take three deep breaths — four seconds in, six seconds out — before you turn to page one. This buys you thirty seconds and calms your system.

Start with a question you know you can do. Not necessarily question one. Skim the paper for a friendly one and start there. Confidence unlocks the rest.

If you get stuck on a question, put a small dot next to it and move on. Come back after you have finished the rest. Sitting stuck loses points on questions you could otherwise get.

If a wave of panic hits mid-paper, put your pen down for ten seconds, breathe, and re-read the last sentence you wrote. Do not stand up. Do not close your eyes for long. Ten seconds and you are back.

Walking counts. A brisk walk after dinner, even in the last month before the exam, does more for your next study session than the extra thirty minutes at the desk would have.

Do not start a new sport in the last two months. This is not the time for a first attempt at basketball or a first long run. Injuries in June are the worst.

Team sports twice a week are useful for mood as much as for fitness — the social contact is a real part of the benefit.

In the last three days before the exam, keep moving gently but do not do anything that could injure you. A twisted ankle on June 6 is not a story you want.

Pack your bag by 8 PM. Admission ticket, ID card, pencils, pens, ruler, compasses, eraser, water bottle. Lay the outfit out for the morning.

Eat a light, familiar dinner. Do not have coffee after 6 PM.

Read something for pleasure for thirty minutes. Not a textbook. Not a mock paper. A novel, a comic, anything that is not the exam.

In bed by 10:30 PM. If you cannot sleep, do not panic — lying quietly is worth most of what sleep gives you. Do not check your phone.

Set two alarms. Trust your preparation. Turn off the light.

The consensus top pair is Peking University (北京大学) and Tsinghua University (清华大学), both in Beijing. Beida leans humanities and social sciences; Tsinghua leans engineering and computer science, though both are broad.

Fudan University (复旦), Shanghai Jiao Tong University (上海交大), and Zhejiang University (浙大) form the next tier, with Zhejiang and Jiao Tong particularly strong in engineering, Fudan in medicine and finance.

Nanjing University (南大), University of Science and Technology of China (中科大, in Hefei), and Harbin Institute of Technology (哈工大) round out the top of the science-and-engineering ranks.

Renmin University (人大) in Beijing dominates in economics, law, and journalism. Beijing Normal (北师大) leads for education.

The 双一流 designations updated in 2022 and again in 2025 give the current official list of elite institutions. A more specialized program at a lower-ranked university is often a better choice than a marginal admission at a top one.

Members: Peking University, Tsinghua University, Fudan University, Shanghai Jiao Tong University, Zhejiang University, Nanjing University, University of Science and Technology of China, Harbin Institute of Technology, and Xi'an Jiaotong University.

Together the nine account for a very small fraction of national university enrollment — under 1% — but disproportionately large shares of research output, national grants, and prominent graduates.

Membership has been stable since 2009, though the more recent 双一流 designations have somewhat blurred the C9's exclusivity by elevating a broader set of universities and disciplines.

For most students, 'C9' is a useful shorthand for 'top-tier research university with national profile' and less useful as a strict admission target — the individual universities and their programs are what actually differ.

Common destinations are the US, UK, Canada, Australia, Hong Kong, and Singapore, with each system requiring different combinations of standardized tests (SAT, ACT, IELTS, TOEFL), high school transcripts, and application essays.

The decision to skip the gaokao is usually made in tenth grade or earlier. Late switches — deciding in twelfth grade that you will apply abroad after all — often result in a rushed, weak application.

It is possible to sit the gaokao and apply abroad simultaneously, and some students do — this is expensive and stressful and works best when the family has been planning both tracks from the start.

Some universities in Hong Kong, Australia, and increasingly a handful in the UK and US will consider a strong gaokao score as part of an application. This creates a hybrid track worth investigating.

But the labor market has softened in some technology sectors, and the graduating cohorts in computer science are now very large. A middling degree from a middling program no longer buys the salary premium it did a decade ago.

Meanwhile, some historically 'low-return' majors — law, medicine, teaching — have quietly improved their long-run outcomes as demand has caught up with supply.

The honest answer to 'what should I major in?' remains: choose a subject you can be good at, in a program that is well-regarded, and worry about the specific job later. Being top of your class in a less-hyped major usually beats being middle of the class in a fashionable one.

Interdisciplinary programs — computational finance, biomedical engineering, computational linguistics — are worth serious attention if your interests and school allow. They tend to be underpriced by cutoff scores relative to their long-run career flexibility.

Orientation week (迎新周) is a strange, packed period. Do not sign up for every club that has a table. Pick two — one that matches an existing interest, one that scares you a little — and let the rest wait until the second semester.

Roommates are assigned. You will probably not choose them. Kindness in the first week pays enormous dividends over four years. So does not judging anyone's home province.

The academic pace is looser than senior year of high school, but the reading load in humanities and the problem sets in engineering catch up with students who mistake looseness for lack of demand. Do the first month's work in the first month.

Homesickness is real and it usually peaks around week three. Call home more than you think you need to. It helps them and it helps you.

You are through the gaokao. The rest of your life is not an exam. Try to remember that.