牛剑本科 中国申请者 录取率 笔试面试 | Oxbridge Undergraduate Admissions: Chinese Applicants, Acceptance Rates, Written Tests and Interviews

牛剑本科 中国申请者 录取率 笔试面试

剑桥大学在 2026 年 5 月发布了 2025 申请季的本科招生统计，中国大陆的申请量达到 2,900 份，入学人数 284 人。把这份表和上一年的并排放在一起，会看到一条很清楚的线：申请涨得比名额快。很多学生以为牛剑的申请池每年都在暴涨，其实总量只比上一年多了一个多百分点，真正的增量集中在海外这一栏。这篇文章把 2025 与 2024 两季的官方数字拆开，说明中国申请者现在面对的是什么结构，以及笔试、面试、选课三件事该怎么排。

Cambridge published its 2025 cycle admissions statistics in May 2026, and the mainland China line is the one to read first: 2,900 applications, 284 acceptances. The totals barely moved year on year, yet the overseas column grew while the home column flattened. What follows is a reading of those tables, plus how the tests and interviews actually decide who gets in.

2025 申请季：22,513 份申请与 3,669 人入学

剑桥官网这份报告的总量是 22,513 份申请、4,893 份 offer、3,669 人最终入学，整体成功率 16.3%。上一轮 2024 申请季是 22,153 份申请、4,760 份 offer、3,632 人入学，成功率 16.4%。两年相比，申请量增加 360 份，涨幅 1.6%，入学人数增加 37 人，涨幅约 1%。总量的稳定掩盖了结构上的移动，本土申请略降，海外申请明显上升。招生官读这张表的方式和申请者不同，名额在申请季开始前就已经定好，申请量的变化只改变竞争密度，不改变名额本身。剑桥按 domicile 划分申请者，海外单独成表，不和本土混算。

Cambridge's 2025 cycle report records 22,513 applications, 4,893 offers and 3,669 acceptances, a success rate of 16.3%, against 22,153, 4,760 and 3,632 a year earlier. The overall pool has hardly moved, which hides a clear shift inside it, as overseas demand rose while home applications flattened. Admissions staff read these tables as a supply story rather than a difficulty story, because the number of places is fixed long before anyone applies.

中国大陆：2,900 份申请，284 人入学

剑桥官网的国别表里，中国大陆递交 2,900 份申请，占海外申请总量的 34.9%，拿到 360 份 offer，最终 284 人入学，成功率 9.8%。同一张表里，新加坡 575 份申请、133 人入学，成功率 23.1%；香港 574 份申请、82 人入学，成功率 14.3%；印度 370 份申请、25 人入学，成功率 6.8%；美国 597 份申请、33 人入学，成功率 5.5%。中国大陆的申请量与 offer 数都是第一，成功率排在中等偏下的位置。招生官看的是同一个池子里的相对位置，不是单独的绝对分数。很多学生以为三个 A* 就有了竞争力，同一年递交三个 A* 的中国申请者数量，远超剑桥能给的名额。

Mainland China supplied 2,900 applications in the 2025 cycle, 34.9% of all overseas applications, and converted them into 360 offers and 284 acceptances, a 9.8% success rate. Singapore and Hong Kong converted far better at 23.1% and 14.3%, while India and the United States sat at 6.8% and 5.5%, which shows how uneven the overseas pool is. Grade profiles alone cannot explain that spread, since subject choice, test scores and interview performance do most of the work once a file reaches the academics.

两年对比：申请涨 25.6%，录取涨 12.7%

把 2024 与 2025 两季的中国大陆数字并排看：申请从 2,308 份涨到 2,900 份，涨幅 25.6%；offer 从 343 份涨到 360 份，涨幅 5.0%；入学从 252 人涨到 284 人，涨幅 12.7%。offer 率从 14.9% 降到 12.4%，成功率从 10.9% 降到 9.8%。同一时期海外整体申请从 7,638 份涨到 8,300 份，涨幅 8.7%，海外成功率从 11.1% 降到 10.7%。中国大陆的申请增速接近海外整体的三倍，这是过去两年最明显的变化。名额没有同步扩张，多出来的申请挤在同一批专业里，压力落在边际申请者身上。常见的错误是只盯着自己那一届的分数线，忽略了同届竞争者的数量与专业分布。

Between the two cycles, mainland applications rose 25.6% while offers grew only 5.0%, so the offer rate fell from 14.9% to 12.4% and the success rate from 10.9% to 9.8%. Overseas applications as a whole grew 8.7%, which means Chinese demand expanded roughly three times faster than the rest of the overseas pool. A one point drop looks small in a table, yet it separates an offer from a rejection for dozens of applicants every year.

海外 10.7% 对本土 19.6%

2025 季剑桥本土申请 14,213 份，3,672 份 offer，2,779 人入学，成功率 19.6%；海外 8,300 份申请，1,221 份 offer，890 人入学，成功率 10.7%。本土的 offer 率是 25.8%，海外是 14.7%，差 11 个百分点。这个差距有一部分来自名额结构，海外名额单独一栏，学费与资助安排也不同。另一部分来自材料的可读性，本土申请者有 A-Level 预测成绩、学校推荐与 widening participation 背景数据，海外申请者的学校背景对招生官来说更难判断。很多学生以为海外身份就是劣势的全部原因，材料读不懂这一层很少被讨论。

Home applicants reached a 19.6% success rate against 10.7% overseas, and the offer rate gap was wider still at 25.8% versus 14.7%. Part of that gap is structural, since overseas places sit in a separate column of the admissions plan with different fee arrangements. The rest comes from how readable a file is, because predicted grades and school context are familiar for a UK school and much harder to calibrate for a school the admissions tutor has never seen.

专业冷热：计算机 7.4%，经济 10.0%，医学 17.6%

2025 季剑桥计算机科学 1,739 份申请、128 人入学，成功率 7.4%；经济 1,647 份申请、164 人入学，10.0%；工程 3,263 份申请、335 人入学，10.3%；数学 2,032 份申请、258 人入学，12.7%；法律 1,654 份申请、232 人入学，14.0%；医学 1,530 份申请、270 人入学，17.6%；自然科学 2,644 份申请、576 人入学，21.8%。计算机从 2024 季的 1,863 份申请降到 1,739 份，工程从 2,654 份涨到 3,263 份，医学从 1,791 份降到 1,530 份。中国申请者集中在数学、工程、经济、计算机四个方向，这几个方向的成功率都在 7% 到 13% 之间。换成自然科学或医学，同样的学术水平面对的成功率完全不是一个量级。选专业时把成功率放进决策，比只看专业排名更接近现实。

Computer Science admitted 128 of 1,739 applicants (7.4%), Economics 164 of 1,647 (10.0%) and Engineering 335 of 3,263 (10.3%), while Natural Sciences reached 21.8% and Medicine 17.6%. Engineering applications jumped by more than 600 year on year and Computer Science fell by about 120, so the crowd moves between subjects rather than leaving the system. Chinese applicants cluster in the four most competitive columns, which is a decision about identity as much as about difficulty.

牛津一侧的口径与考试安排

牛津官网每年发布本科招生统计报告，分专业、分国籍列出申请、offer 与入学人数，口径与剑桥并不一致，跨校比较需要先读注释。牛津的筛选更早进入专业内容，多数专业在面试前安排学科笔试，法律考 LNAT，医学考 UCAT，数学与物理类专业的笔试名称近两年有过调整，报名前以牛津官网当年的 admissions tests 页面为准。面试由学院安排，通常两到三场，每场两位面试官，题目从 A-Level 大纲的边缘往外延伸。录取决定在学院层面做出，学科系提供学术评估，两个环节看的材料并不相同。牛剑的差别不在难度，在筛选发生的位置，剑桥更依赖笔试分数进入面试，牛津更依赖面试把笔试过关的人再分一层。

Oxford publishes its own undergraduate admissions statistics each year, broken down by course and nationality, and its definitions do not match Cambridge's, so any cross comparison needs the footnotes. Its filter starts earlier inside the subject, with a discipline-specific test before interview for most courses, LNAT for Law and UCAT for Medicine, while interview panels are run by the colleges. The real difference between the two universities is where the cut happens, since Cambridge leans on test scores to build the interview list and Oxford uses the interview to separate candidates who already cleared the test.

分数之外：选课组合与学科深度

剑桥的专业页把入学要求写得很具体，A-Level 通常要求 A*A*A 到 A*AA，科目组合限定在指定清单内，工程要求数学与物理，自然科学要求数学与至少一门科学。中国学生常把高数、物理、化学、经济四门一起考，用来证明广度，招生官看的却是与专业相关的那两三门有没有到最深的一层。很多学生以为多考一门能加分，超出要求的科目只在个人陈述与面试里提供谈资，不改变学术评估的主体。常见的错误是把 Further Maths 当成数学专业的加分项，在剑桥数学的评估里它更接近默认配置。选课的顺序是先定专业，再倒推必考科目，最后才考虑第五门。

Course pages state requirements as a range, typically A*A*A to A*AA at A Level, with named subjects that must appear in the combination, such as Maths and Physics for Engineering. A fourth or fifth subject impresses nobody on its own, because what matters is depth in the two or three subjects the course actually uses. Further Maths is treated as a default rather than a bonus for Cambridge Mathematics, and applicants who add it late often struggle with the step up in the first term.

笔试：ESAT、TMUA 与专业匹配

剑桥从 2025 年入学起用 ESAT 取代原来的 NSAA 与 ENGAA，工程、化学工程、自然科学、兽医等专业考这一场，内容分数学、物理、化学、生物几个模块，按专业选考其中若干。计算机、经济、数学考 TMUA，两卷都是选择题，重点在数学推理的速度与准确度。法律考 LNAT，医学考 UCAT，这两场与剑桥自己的学科笔试分开报名。考试在十月中旬，报名截止通常比 UCAS 早一到两周，中国大陆的考点名额很快满。很多学生以为笔试只是走流程，它其实决定能否进面试，分数段直接对应面试名单。常见的错误是拿 A-Level 真题练手，ESAT 的题型与 A-Level 不完全对应，需要按剑桥官网给的样题与 specification 对齐。

Cambridge uses ESAT for Engineering, Chemical Engineering, Natural Sciences and Veterinary Medicine, with module choices that follow the course, and TMUA for Computer Science, Economics and Mathematics. Both are sat in mid-October, and registration closes earlier than the UCAS deadline, which catches out applicants who treat the test as an afterthought. Past A Level papers help with content but mislead on format, because the sample papers and specifications on the admissions site define what the questions actually look like.

面试：考的是可教性

剑桥的面试通常一到两场，每场二十到四十分钟，面试官是学院里教这门课的人。题目从熟悉的知识点起步，很快走到没学过的情境，看的是拿到提示之后能不能往前走。很多学生以为面试要展示知识量，面试官更在意被纠正时的反应，能不能听懂提示、改掉错误、把思路讲清楚。评分表上没有标准答案，只有几个观察维度，包括能否把问题拆开、能否用例子检验猜想、能否在压力下保持推理连贯。常见的错误是把不会的题硬答成会的题，绕开提示自己讲一套。面试前把专业相关的三四个核心概念练到能自由推导，比刷一百道竞赛题更接近面试的真实要求。

Interviews last twenty to forty minutes and are run by the academics who teach the course, and the questions move quickly from familiar material into situations the applicant has not met. What the panel records is how a candidate responds to a hint, whether the correction is absorbed and whether the reasoning stays coherent under pressure. Candidates who bluff past a prompt usually lose the room, while a slow, honest attempt that improves often leaves a better impression than a polished answer to a different question.

材料链：成绩单与 My Cambridge Application

剑桥官网把 My Cambridge Application 写成申请的必需部分，2025 季有 289 名申请者因为没有提交它而被排除在统计之外。这份表要填课程模块、考试成绩、学校背景与资金安排，填错科目代码会让学术评估组拿不到正确的课程信息。成绩单覆盖最近三年的完整记录，中国学校的中英文对照件需要学校盖章，扫描件清晰度不够会被要求重交。推荐信由学校 counselor 或学科老师写，剑桥想看的是课堂表现与学术潜力，不是形容词的堆叠。很多学生把精力放在个人陈述的措辞上，材料链上任何一环缺失，前面写得再好也进不了面试名单。

Cambridge treats My Cambridge Application as part of the application, and 289 applicants were excluded from the 2025 cycle statistics for not submitting it. The form carries module-level detail, exam results, school context and fee status, so a wrong subject code can leave the academic assessors without the information they need. Transcripts must cover three full years with certified translations, and a blurred scan triggers a request for a resubmission rather than a decision.

三个容易被忽略的判断

申请策略里最容易出错的地方，通常不是成绩不够，而是判断的基准选错了。有的学生拿学校内部排名当参照，忽略了同校往年进入面试的人数；有的把学长学姐的 offer 当成模板，没注意那一届的专业与学院完全不同；还有的只看总录取率，没看目标专业的申请池大小。这三种判断都指向同一个问题，用平均值代替了自己所处的那一栏。剑桥的统计表按专业、国籍、学校类型分栏，找到自己那一栏的数字，才知道剑桥官网给出的 9.8% 整体成功率对个人意味着什么。成功率低不等于不该申请，它只说明笔试、面试、选课三件事需要同时做到位，不能把希望押在其中一项上。

Most strategy errors come from using the wrong benchmark: an internal school rank, a senior's offer from a different course, or an overall success rate averaged across wildly different pools. Cambridge publishes its tables by subject, domicile and school type for exactly this reason, and the relevant column is usually far from the headline figure. A low success rate is not an argument against applying, it is an argument for getting the test, the interview and the subject choice right at the same time.

从现在到十月十五日的时间表

2027 年入学的 UCAS 截止日是 2026 年 10 月 15 日，下一轮按 UCAS 的日历落在十月中旬，准备 2028 年入学的人从现在算起还有十一个月。前三个月确定专业与学院，读官网的 course page 与 specification，确认必考科目与笔试清单。中间四个月把 A-Level 内容学完一轮，用官网样题做笔试训练，记录每次的正确率。最后四个月准备面试与个人陈述，找学科老师做两次模拟面试，把不会的题按提示重做一遍。申请材料在九月前完成初稿，十月留出两周处理 My Cambridge Application 与推荐信。时间表宽不宽松，取决于笔试准备有没有提前，不取决于文书写得多早。

The UCAS deadline for 2027 entry was 15 October 2026, and the next cycle follows the same mid-October pattern, which leaves about eleven months for anyone targeting 2028 entry. Split that window into course selection, subject mastery with test practice, and interview preparation, and keep the paperwork for September rather than October. Applications fail on sequencing far more often than on talent, and the sequence is published on the admissions website months in advance.

需要辅导？

16621398022同微信

关注公众号 tutorhao，更多学习资源见 aleveler.com

更多咨询请联系16621398022（同微信）