Standardised tests for US college applications: which to sit, when, and whether to submit

Standardised tests for US college applications are a planning decision before they are a preparation problem: which tests to sit, when to sit them, and whether to send the result. Three different questions hide inside the one phrase. The SAT and the ACT are general tests taken outside school, accepted interchangeably wherever a test is accepted at all, and the only ones a university may tell you to leave out. The TOEFL and the IELTS answer something else entirely — whether a student can study in English. AP, IB and A-Level results are subject evidence produced by the curriculum the student is already sitting in. Each travels on its own calendar, each can be absent for a different reason, and only the first is ever genuinely optional. Across the twelve admission cases published on this site, an English-language score appears in all twelve records and an SAT or ACT score in ten; two records state the tests as None. This guide defines the terms, sets out what each test can and cannot establish, shows what our own records actually list, and gives the order the decisions are made in. We do not sell test preparation, so none of this is coaching advice. It is the decision that sits above the coaching.

Key definitions

• Admissions test. The SAT or the ACT: a general test of reasoning and attainment, taken outside the school curriculum, on dates the test maker publishes for the whole year. Where a university accepts a test, it accepts either one.
• English-language test. The TOEFL or the IELTS. It establishes that a student can follow instruction in English. It is not an academic-aptitude measure and does not substitute for one.
• Subject result. An externally graded result in a named subject: an AP score, an IB final or predicted grade, an A-Level certificate or predicted grade. It corroborates depth in a subject rather than general ability.
• Predicted grade. The grade a school expects a student to achieve in a curriculum whose final exams fall after the application deadline, as in the IB and A-Level systems. It is the figure that travels with the application.
• Test policy. A university's rule for the application year on whether an admissions-test score is required, optional, or not considered. It is set per university and per year, and the only reliable place to read it is that university's own admissions page.
• Academic depth. In the Visible Growth system at Mellin International Education, the first cornerstone of the method: building and presenting depth in a core interest, so that the file solves the problem of scores without an academic identity. It is also what has to do the work when a score is missing.

What each test can establish, and what it cannot

Test	What it establishes	What it cannot do	When it is fixed
SAT or ACT	A comparable general measure across schools and curricula	Say anything about depth in a subject, or about what the student has produced	Sittings run through the year; the last useful one is set by the application deadline
TOEFL or IELTS	That instruction in English is not a barrier	Stand in for academic evidence of any kind	Valid for a fixed period, so it is sat late enough to still be current
AP	Externally graded subject evidence, subject by subject	Cover a subject the student did not sit	Exams sit in May, so a Grade 11 sitting is the last to arrive before the application
IB or A-Level	Subject evidence from the curriculum already being studied	Be finalised before the application in most cases	The predicted grade travels with the application; the final grade arrives afterwards

Read the third column downwards and the division of labour is clear. Only one row in that table is ever described as optional, and it is the row that establishes the least about the student. That asymmetry is the whole of the submission decision.

What our own records list

Below are the test results recorded in the twelve detailed cases on this site, exactly as the records state them. Nothing here explains why any student was admitted, and none of it is a benchmark; it is what the files contain.

Case	Curriculum as recorded	English test	Admissions test	Subject results as recorded
Caltech and Oxford, physics	A-Level	TOEFL 116	SAT 1540	Not stated
Cornell, information science	IB	TOEFL 112	SAT 1550	Not stated
UPenn, arts and sciences	Not stated	TOEFL 115	SAT 1570	Not stated
Cornell, astronomy	A-Level	TOEFL 112	None	Not stated
Duke and Rice, biology	IB	TOEFL 115	SAT 1550	Not stated
Duke, CMU and UCLA, computer science	AP	TOEFL 110+	SAT 1500+	AP calculus BC, computer science A, world history, English language and composition, US history, biology — all 5
Emory, neuroscience	IB	TOEFL 115	SAT 1580	IB predicted 37 / 42
Brown, gender studies and history	Own curriculum, AP	TOEFL 115+	SAT 1550+	AP US history 5, calculus BC 5, statistics 5, English language and composition 4
Johns Hopkins, East Asian studies	AP	TOEFL 115 (reading 28, speaking 29, listening 30, writing 28)	SAT 1550 (maths 760, reading and writing 790)	AP calculus BC 5, calculus AB 5, microeconomics 5, psychology 5, statistics 5, English literature and composition 4, plus US history and world history sat in May 2024
Harvard, mathematics	Public high school international division	TOEFL 110+	ACT 36	Not stated
Carnegie Mellon, mechanical engineering	AP	TOEFL 111	None	AP, four 5s and one 4
Tufts, biotechnology	A-Level	TOEFL 116	SAT 1570	A-Level biology, chemistry, further mathematics, history, English language; predicted A* in biology and chemistry, A in the rest

The shorter admission highlights on the same page add three more data points that the table above does not contain: one student is recorded with A-Level 4A* and IELTS 7.5, one with IB 40 and TOEFL 105, and one with GPA 4.28 and ACT 35.

Five readings follow, and all five are observations about what the records contain.

1. The English-language score is the constant; the admissions test is not. All twelve detailed records carry a TOEFL result, in a band from 110+ to 116. Ten carry an SAT or ACT. If one of the two had to be dropped from a plan, the records are unambiguous about which one the file can survive without.
2. Two admits are recorded with no SAT or ACT at all. The astronomy record states the top 15% of the year and None; the mechanical engineering record states the top 30% and None. Both sat a TOEFL. Neither is a story about a missing score; both are stories about what was there instead.
3. One record in twelve uses the ACT, and one highlight uses the IELTS. The site's record is overwhelmingly SAT and TOEFL. That reflects what the students in front of us chose rather than any claim about what works better.
4. Where the record is precise, it is very precise. The Johns Hopkins entry lists four TOEFL section scores and both SAT sections, eight AP subjects, and two of them as a May sitting rather than a score. That is what a test record looks like when it is kept rather than reconstructed.
5. Subject results appear in six of the twelve, and they track the field applied for. Both humanities records carry calculus and statistics. The biotechnology record carries predicted A* grades in the two sciences the application was built on. A subject result is only corroboration if it corroborates something, which is a point our academic file guide makes in detail.

Deciding whether to submit

The decision is file-specific, and it runs in one direction: a score corroborates an argument, and never makes one. Four questions, in order.

1. What does this university say, for this application year? Policies range from required to optional to not considered, they are set university by university, and they change. Read each one on that university's own admissions page and write the answer next to the school on the list. Nothing else in this section matters until that column is filled in.
2. What is the rest of the file already establishing? If the transcript is strong and the evidence is specific — a paper, a built thing, a programme with participants — a score adds a line to a case that is already made. If the academic direction is still thin, or the transcript sits on a scale an admissions reader will not recognise, a strong score is cheap corroboration and worth sending.
3. Does the score point the same way as the file? A quantitative application with a weak maths section, or a humanities application with a weak reading and writing section, sends a score that argues against the file. Our highly selective applications guide sets out the general rule this is an instance of: every document has to say the same thing about the same person.
4. What does the decision do to the list? A file with no admissions-test score is not a weaker file, but it is a differently shaped one, and the shape belongs in the list rather than in the autumn. The reach, match and safety guide is where that gets settled, and the round decision follows the list rather than leading it.

When to sit, re-sit and stop

1. Put the English test first and early. It is the one result every record on this site carries, it is valid for a fixed period rather than forever, and it is the least strategic decision in the set. Sit it in Grade 11, confirm it is still current at the application deadline, and take it off the table.
2. Choose one admissions test, once. Sit a full-length timed version of each under real conditions, compare the results, commit to the one the student actually finishes, and never split preparation across both.
3. Work backwards from the deadline, not forwards from the next available date. Count the sittings that fall before the earliest application deadline on the list, then plan the first attempt early enough to leave one re-sit inside that window.
4. Settle the predicted grades in Grade 11. In the IB and A-Level systems, the figure that travels with the application is the prediction, not the final result, so the conversation with the school happens a year before the deadline. Our senior planning mentor works across the A-Level, IB and AP systems and reverse-engineers the planning path from the admitted-student profiles of target universities, which is the same calendar read from the other end.
5. Treat the May AP sitting as a boundary. A Grade 11 sitting arrives before the application; a Grade 12 sitting does not. Decide which subjects need to be on the record by the deadline and which can wait.
6. Stop on a plan, not on a feeling. Decide in advance how many attempts the plan allows and what score would end it. Two more months spent on a third attempt is two months not spent on the thing in the next section.

What carries the weight when there is no score

Both of our records with no admissions test replaced it with something that exists. The astronomy file carries three papers, a round-table seminar that grew from four people to over a hundred, and an online astronomy course built for 89 children in Gansu after the student was told their lack of prior knowledge made it impossible. The mechanical engineering file carries a tilt-rotor drone design with an 80-page report, first authorship of an IEEE paper from research at Tsinghua, and an aviation club of 800 members across four cities. That is the practical content of our first cornerstone: depth built in a core interest and then presented, so the file is not asking a score to speak for it.

This is also why a research placement is planned for its output rather than its name. Our research and programme work connects students to laboratories and researchers at leading institutions, and the point of a placement is the recommendation letter or academic report that comes out of it — evidence of a core academic interest, which is exactly the thing a missing number cannot supply and a present number cannot either. Several people on our mentor team have read application files from the other side of the desk, and the consistent observation is that the number is the quickest part of a file to read and the least informative.

What we recommend

Fill in the policy column for every university on the list before discussing scores at all, then decide per file rather than per rule. Sit the English test in Grade 11 and stop thinking about it. Choose one admissions test on evidence from a timed attempt, plan the sittings backwards from the earliest deadline with one re-sit inside the window, and set the stopping condition before the first attempt rather than after the second. Settle predicted grades with the school a year early. Submit a score when it points the same way as the rest of the file and withhold it when it does not, and make that decision visible in the school list instead of in the final fortnight. Above all, do not let the test calendar eat the year that was supposed to produce the evidence: our long plans are broken into stage goals and weekly tasks precisely so that both fit, and the three-year roadmap shows where each belongs. If you want a current file read for whether a score helps or hurts it, tell us the student's grade, curriculum and interests and a consultant will reply within one working day.