How the scoring works.

Your recording is transcribed, and the transcript is scored out of 10 on each of 8 competencies: Communication, Empathy, Ethical Reasoning, Professionalism, Structure, Leadership, Critical Thinking, Confidence. Interview Readiness is the plain mean of those, not a separate weighted metric, so it can always be explained by the numbers underneath it.

No Australian medical school publishes its MMI marking rubric. This is the most important limitation on the site and it applies to the product it is named after.

So these eight competencies are not a school’s rubric, and cannot be. They are assembled from the qualities university admissions pages state they are assessing, and from published medical interview frameworks. That is a weaker claim than “marked against the published rubric”, and it is the true one. Copy on this site that said otherwise was wrong and has been corrected.

What it cannot see

• Anything that is not in the words. Eye contact, posture, warmth in the voice, and the pauses a real examiner reads a great deal into.
• Transcription error. An accent, a mumble or background noise becomes a word the marker then scores, and it has no way to know the word was not the one you said.
• Whether a real panel would have liked you. Two human examiners disagree with each other; a model agrees with itself, which is not the same as being right.

When there is nothing to score - a silent recording, an answer that never arrived - the marker says so rather than producing a number. A score of zero and an absence of evidence are different things.

The one score here that is marked against a genuinely published standard. ACER publishes what it assesses Written Communication on, and essays are marked against 5 criteria derived from it: Thesis and Argument, Structure and Cohesion, Evidence and Examples, Expression and Language, Depth and Insight. Each is scored out of 10, with the reasoning shown per criterion.

What it cannot see

• How ACER's own markers would have banded it. The criteria are published; the marking guide, the exemplars and the moderation are not. The band is an estimate against the criteria, not a prediction of your result.
• The rest of the cohort. GAMSAT reports a scaled score relative to everyone who sat it, and that distribution is not knowable in advance.

Sets are marked, scaled to the 300 to 900 range UCAT ANZ reports, and the three cognitive subtests are summed to a 900 to 2700 total. Situational Judgement is scaled the same way and reported separately, because that is what UCAT ANZ does - the Band 1 to Band 4 system belongs to UCAT UK and would be a number from the wrong country’s exam.

The percentile comes from the official 2025 scaled-score distribution: the 16,950 candidates who tested between 1 July and 17 August 2025. UCAT publishes that as deciles and quartiles, not as a formula, so a percentile here is a straight-line interpolation between the published anchors - exactly how you would read the table by eye.

Where the data stops

• Below the 10th or above the 90th percentile the published table ends. The estimate reports 'bottom 10%' or 'top 10%' rather than inventing a precise figure the source cannot support.
• It is last year's cohort. Difficulty and scaling move between years, and a percentile against 2025 is a guide to 2026, not a measurement of it.
• A practice set is not a test-day sitting. Nerves, fatigue and a two-hour continuous exam are not reproduced by a short set.

Your GPA and GAMSAT are placed against applicants who reported an offer, and against those who reported a rejection, at each university with enough reported outcomes to say anything. The current data holds 265 reported offers and 186 reported rejections.

This is community-sourced and self-reported. It is not GEMSAS or ACER data, and no university has supplied or endorsed it.

What it cannot see

• Preference order, rural or Indigenous entry pathways, portfolio, and interview performance - several of which decide a marginal case outright.
• How strong a given year's field was. A cutoff moves with the applicant pool.
• Who did not report. People who are disappointed and people who are delighted do not fill in a spreadsheet at the same rate, and that skews the sample in a direction nobody can measure.

Where a cohort is too small to support an estimate, the calculator says so rather than producing a percentage that looks as solid as one built on hundreds of outcomes. Always check requirements against the university’s own admissions page.

Practice feedback, not an admissions decision. Nothing here predicts your result, no examiner or university has endorsed it, and this site is not affiliated with ACER, GEMSAS, or any university.

The purpose of a score here is to tell you where to spend your next hour. That is a job an estimate can do honestly, and it is the only job these numbers are built for.