How Predicted Grades Are Set for the 2026-27 UCAS Cycle

A UCAS predicted grade is the grade a school or college believes an applicant is likely to achieve in positive circumstances, set by the school itself rather than by UCAS or an exam board. Around half of UK 18-year-olds were predicted at least AAA at A-level for 2025 entry, yet only 26% of accepted applicants actually achieved it. The prediction is set once, submitted with the UCAS application, and, in almost every case, cannot be changed afterwards.

Who Actually Sets a Predicted Grade, and What Schools Are Required to Do

UCAS does not set a predicted grade, and neither does an exam board. The school or college where a student is studying sets it, entered by a subject teacher and submitted through the institution’s own UCAS adviser account, most often once a student begins studying towards a level-3 qualification such as an A-level. UCAS’s own guidance is explicit about the obligation this places on a school: “As a registered centre, you’re responsible for ensuring predicted grades and a reference are added to every application.” A registered centre cannot send an application for a pending qualification without either entering a predicted grade or acknowledging a warning that one is missing.

UCAS asks schools to follow five stated principles when setting a prediction. A grade should be achievable, “based on demonstrable evidence of performance,” rather than simply aspirational. It should be objective and data-driven, drawing on “past performance, internal assessments, or current student work.” It should be comprehensive, covering every pending qualification unless the qualification’s own structure rules that out. And critically, UCAS states a prediction should be “determined by professional judgement” and “set independently of an applicant’s university or college choices” — a teacher is not supposed to inflate a prediction because a student is aiming for a demanding course, even though universities and colleges then “place differing levels of importance on predicted grades and use them in a variety of ways.”

What UCAS does not specify is how a school arrives at that judgement in practice, and it says so itself. Suzie Rees, UCAS’s own Regional Adviser Lead, put it plainly: “Every school or college might approach this slightly differently, but having a transparent policy really helps everyone stay informed.” That is worth taking at face value rather than smoothing over. There is no single national process behind the number that ends up on an application, only a shared set of principles that leaves real room for one school’s Year 13 team to weigh mock results more heavily than another’s, or to lean harder on coursework where a subject allows it.

Why International Applicants Often Have No Predicted Grade At All

A predicted grade, as UCAS defines it, only exists for a qualification a UK school or college is currently teaching towards. UCAS says so directly, in guidance aimed at applicants outside the UK: “The UK undergraduate application process is different to how some countries manage applications for higher education. In the UK your school or college can provide predicted grades for qualifications you’re studying now that lead to higher education. That way you can apply earlier in the year to increase your chances of getting a place.” For an applicant sitting a qualification with no equivalent forecasting step — many national school-leaving systems are graded once, at the end, with no formal prediction issued beforehand — there is simply nothing for a UK school to enter, because no UK school is teaching them.

This matters for how an international application is actually built. UCAS’s guidance for international advisers describes the mechanism the same way it describes it for UK students — “UCAS predicted grades are the academic scores that an applicant’s school or college expects them to achieve in normal (or ‘positive’) circumstances,” used to decide whether to offer “a place on a provisional basis” — but that description only applies where a predicted grade exists at all. Where it does not, a university has to build its provisional offer from whatever the applicant’s own system produces instead: a school reference, recent transcripts, standardised test scores, or a formal comparison of the qualification against UK equivalents. UCAS points applicants toward UK ENIC, the UK’s National Information Centre, for exactly that — a formal “statement about how your qualifications compare to UK equivalents.”

The two published accuracy studies referenced later on this page make the same point from the data side rather than the guidance side. The most detailed academic analysis of UK predicted-grade accuracy, Murphy and Wyness’s 2020 study for UCL’s Centre for Education Policy and Equalising Opportunities, is built entirely from UK domestic admissions data, and the authors are explicit that it goes no further: “Nor does our study include students applying to UK colleges from the EU or overseas.” The mechanism this page describes, and the research measuring how accurate it is, both stop at the UK system’s edge. An applicant moving into that system from outside it is not missing out on a more accurate prediction — they are outside a process that, for their qualification, was never designed to apply to them in the first place, and their application is built differently as a result, weighted instead toward the evidence their own system does produce.

When Are Predicted Grades Set and Submitted in the 2026-27 Cycle?

Predicted grades for 2026-27 applications follow the UCAS calendar rather than a single fixed date that applies to every student equally. UCAS opened its search tool for 2027-entry courses on 28 April 2026, and applications opened for registration on 12 May 2026. Completed applications, each one carrying its predicted grades, could first be submitted to UCAS from 1 September 2026.

Two deadlines then apply, depending on the course. Applications to Oxford, Cambridge, and to most medicine, dentistry and veterinary medicine or science courses must arrive by 18:00 UK time on 15 October 2026 — UCAS states this is the “equal consideration date for applications to the universities of Oxford and Cambridge, and for most courses in medicine, dentistry, and veterinary medicine/science.” Every other course on the standard timetable follows the main equal-consideration deadline: 18:00 UK time on 13 January 2027.

Because a predicted grade normally cannot be changed once an application is sent, the date a school actually finalises its predictions falls somewhere inside this window, not on it — a school preparing an October submission has far less time to gather mock results and coursework evidence than one working to the January date. Applications that miss both dates are not shut out entirely: UCAS will still forward applications received by 18:00 UK time on 30 June 2027 to universities and colleges, after which any new application is entered automatically into Clearing rather than the standard cycle. These dates were checked live against UCAS’s own published 2026-27 application timetable on 23 August 2026.

How Accurate Are Predicted Grades in 2026? Start With the Current Data

The two most recent, UK-wide figures both come from UCAS and Ofqual directly, and both describe the same shape: a wide gap between what students are predicted and what they go on to achieve, concentrated at the top grades. UCAS’s own current guidance states it plainly for 2025: “around half of UK 18-year-olds predicted AAA and above in 2025. In comparison 26% of accepted applicants secured AAA and above last summer.” Half of the country’s strongest applicants were told they were on course for three A grades; roughly a quarter of accepted applicants actually left the summer holding them.

The second current figure comes from a joint letter Ofqual and UCAS sent directly to students ahead of results day, on 16 July 2024, describing the cycle that had just finished: “Last year, about one in five accepted applicants achieved or exceeded their predicted grades. Yet over 80% of UK 18-year-olds applying to UCAS secured a place at university or college.” Read together, the practical point is the one UCAS itself is making: falling short of a prediction is the outcome for roughly four accepted applicants in five, and it still is not the same as missing out on a place. The same letter notes that a record 39,620 UK 18-year-olds secured a place through Clearing in 2023 alone.

Both figures are genuinely current — 2025, and the 2023 cycle respectively — and both are UCAS’s or Ofqual’s own numbers, not a third party’s estimate of them. For more mechanism behind why the numbers land where they do, two older studies go into far more depth: a 2011 government-commissioned report examining the 2009 A-level cohort, and a 2020 peer-reviewed academic study covering 2013 to 2015. Both predate the 2015-2020 move to linear, exam-only A-levels and the pandemic-era disruption to how grades were subsequently awarded, so they are read below as evidence of how the prediction mechanism behaves, not as this cycle’s own numbers.

Why a 51.7% Grade Accuracy Rate and a 16% Applicant Accuracy Rate Are Not a Contradiction

The two older studies appear, at first glance, to disagree sharply. A 2011 report for the Department for Business, Innovation and Skills — written by UCAS’s own Nick Everett and Joanna Papageorgiou, examining 219,744 achieved A-level grades from 97,268 applicants in the 2009 admissions cycle, about 30% of all UK-domiciled A-levels taken that year — found that 51.7% of all predictions were accurate. A 2020 study by Richard Murphy and Gill Wyness, published through UCL’s Centre for Education Policy and Equalising Opportunities using UCAS admissions data from 2013 to 2015, found that “only 16% of applicants achieve the A-level grade points that they were predicted to achieve by their teachers.” One figure is more than three times the other, measured from data supplied by the same organisation.

They are not measuring the same thing, and the difference is stated in each paper’s own wording. The 2011 figure counts individual predictions: one grade, in one subject, checked against what the student actually achieved in that subject. The 2020 figure counts whole applicants, and specifically asks whether a student’s best three A-levels, taken together, landed exactly on what all three were predicted to be. Getting one subject right is a very different event from getting three subjects right at once.

If a single grade prediction is accurate 51.7% of the time, and a student’s three A-level predictions behaved as three independent events, the chance of all three landing exactly right would be 0.517 × 0.517 × 0.517 — a figure we have calculated ourselves, not one published by either study, and it comes to 13.8%. That is close to Murphy and Wyness’s measured 16%, and closer still to the Ofqual/UCAS letter’s “about one in five” figure from the far more recent 2023 cohort, cited above. None of this makes the numbers interchangeable: our 13.8% assumes three predictions behave as independent events, which they will not exactly, since a student’s subjects, and a school’s approach to predicting them, tend to move together rather than separately, and a strong or weak Year 13 typically shows up across a whole set of predictions rather than in just one. Treat the multiplication as an illustration of why the applicant-level number is so much lower than the grade-level number, not as a fourth published study, and treat the closeness between 13.8%, 16% and “about one in five” as a sign the two headline figures describe the same underlying pattern rather than a genuine disagreement between UCAS’s data and UCL’s analysis of it.

Why Accuracy Varies by Grade, School Type and Background

The BIS report’s 51.7% figure is a per-grade-prediction average across its entire 2009 sample, and it moves a long way depending on what was actually predicted. A-grade predictions were the most reliable that year: 63.8% of predicted A grades were achieved exactly. C-grade predictions were the least reliable, at only 39.4%. The same 2009-cohort dataset also breaks accuracy down by the type of centre a student attended: independent schools reached 64.7% overall accuracy and 73.4% accuracy specifically on A-grade predictions, against 40.2% overall and 49.3% on A grades for further/higher education centres — a gap the report itself partly attributes to independent schools making a much higher share of their predictions at grade A in the first place, where accuracy is naturally higher.

The 2013-2015, applicant-level Murphy and Wyness study finds a different kind of variation, in who a prediction favours rather than in how reliable it is. Among students who ultimately achieved identically high results, the study found that “those from disadvantaged backgrounds receive predicted grades that are lower than those from more advantaged backgrounds” — two equally strong students, on similar evidence, ending up with different predictions depending on where they started from. Neither figure set is this cycle’s own number: both are historical, one from before the 2015-2020 move to linear A-levels and one ending five years before that reform completed. What they show, taken together, is not that predictions were unreliable in some uniform way, but that the reliability itself was — and, on the school-type and equity gaps at least, plausibly still is — uneven across grades, institutions and backgrounds, in ways a single national accuracy percentage cannot show.

Can a Predicted Grade Be Corrected, and Who Can Actually Ask?

A student does not see their own predicted grade by default. UCAS’s own adviser-portal guidance states this without qualification: “Your students won’t see their reference or predicted grades.” Whatever a family knows about a prediction before it is submitted comes from the school choosing to share it, not from any UCAS mechanism that shows it to the applicant directly.

Before an application is sent, a predicted grade can be changed, but only through a specific step: the school’s reference has to be “unapproved” first. UCAS’s guidance is direct about this — “A reference must be unapproved if you want to amend the predicted grades” — and once a reference is re-approved and the application is sent, that window closes. UCAS states the rule for what happens next without hedging: “Once an application has been sent to UCAS we cannot return it. UCAS cannot make changes to the reference or predicted grades once the application has been sent.”

There is one narrow exception, and UCAS is careful to describe it as limited rather than guaranteed. Genuine data-entry errors can sometimes still be corrected after submission, but, in UCAS’s own words, “this may not always be possible for highly selective courses, and reconsideration of an application could be subject to a course still being open, or places still being available.” A wrongly-entered grade on an application to a course with hundreds of applicants for a handful of places is a very different proposition from the same error on a course still recruiting weeks later — the correction depends on the receiving course’s own position at the moment someone asks, not on UCAS’s willingness to fix its own record.

So the honest answer to “can it be changed” has two very different periods in it: freely, before submission, through the school’s own reference-editing process; and only exceptionally, and never guaranteed, after. A family who disagrees with a prediction has real leverage only in the first period, which is exactly why UCAS’s own advice to schools is to make the process “transparent and available as early as possible.”

What Happens If the Final Grade Doesn’t Match the Prediction

This page is about how a prediction is set and how reliable it turns out to be, not about what to do once results are actually in — our guide to what a specific grade boundary actually clears covers that separately, in full. Two figures already given above are worth repeating here because they set the right expectation before results day arrives: only around one in five accepted applicants match or beat their prediction in a typical recent cycle, and 39,620 UK 18-year-olds still secured a place through Clearing in 2023 alone, a record year. Falling short of a predicted grade is the ordinary outcome for most accepted applicants, not an unusual one, and it is not the same as losing a place.

What actually happens next depends on the specific offer, the specific course, and how far short the final grades fall — universities confirm places against the grades a student actually achieves, not the ones predicted for them, so a changed outcome on results day changes what an offer requires, not whether the process itself has failed. That detail belongs to results day itself, not to this page.