Data story · TARA · 2026-entry candidates, first technical report

In TARA's first year, the middle candidate scored 4.0 in Critical Thinking and 4.7 in Problem Solving; the top tenth started at 6.7 and 7.0

UAT-UK's first technical report covers all 2,626 candidates for 2026 entry. It is the only published benchmark before Oxford's first TARA cohort sits in October 2026.

  • By TARAGenius
  • Published
  • Data checked
  • Cite this
4.0 / 4.7Median score in Critical Thinking / Problem Solving, all 2,626 candidates (scale 1.0 to 9.0)Official data
6.7 / 7.0Score at the 90th percentile, Critical Thinking / Problem SolvingOfficial data
742 of 2,626Candidates who left a Problem Solving question unreached or answered it in 5 seconds or lessOfficial data

TARA scores by module and sitting, 2026 entry

Bar: middle half of candidates (25th to 75th percentile). Dot: median. Tick: 90th percentile. Scale 1.0 to 9.0 per module.

Data (CSV)
  • 25th to 75th percentile
  • Median
  • 90th percentile
Critical ThinkingAll candidates (2,626)Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.7Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.7Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.74.06.7October 2025 (976)Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.0Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.0Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.04.57.0January 2026 (1,650)Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.6Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.6Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.64.06.6Problem SolvingAll candidates (2,626)Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.0Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.0Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.04.77.0October 2025 (976)Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.0Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.0Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.04.57.0January 2026 (1,650)Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.0Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.0Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.04.87.01.02.03.04.05.06.07.08.09.0Scaled score, 1.0 to 9.0 per moduleCritical ThinkingAll candidates (2,626)Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.7Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.7Critical Thinking, All candidates (2,626): 25th 2.8, median 4.0, 75th 5.4, 90th 6.74.06.7October 2025 (976)Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.0Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.0Critical Thinking, October 2025 (976): 25th 2.9, median 4.5, 75th 5.8, 90th 7.04.57.0January 2026 (1,650)Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.6Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.6Critical Thinking, January 2026 (1,650): 25th 2.8, median 4.0, 75th 5.2, 90th 6.64.06.6Problem SolvingAll candidates (2,626)Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.0Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.0Problem Solving, All candidates (2,626): 25th 3.5, median 4.7, 75th 5.8, 90th 7.04.77.0October 2025 (976)Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.0Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.0Problem Solving, October 2025 (976): 25th 3.5, median 4.5, 75th 5.8, 90th 7.04.57.0January 2026 (1,650)Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.0Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.0Problem Solving, January 2026 (1,650): 25th 3.5, median 4.8, 75th 5.7, 90th 7.04.87.01.02.03.04.05.06.07.08.09.0Scaled score, 1.0 to 9.0 per module

TARA; 2026 entry (October 2025 and January 2026 sittings); 2,626 candidates; scaled score 1.0 to 9.0; published quantiles, UAT-UK TARA Technical Report 2025-26 (Table 4, page 9); data version 2026-09-26.1

What the data shows

How TARA is scored

Each TARA module is scored from 1.0 to 9.0, and there is no official combined score. The scale is fixed on the October sitting: each cycle sets October's median at 4.5 and its 90th percentile at 7.0, so a 4.5 means "about the middle of the October group" by design.

The first benchmark

Across all 2,626 candidates, the median was 4.0 in Critical Thinking and 4.7 in Problem Solving. The middle half scored 2.8 to 5.4 and 3.5 to 5.8. A 6.7 in Critical Thinking or a 7.0 in Problem Solving put a candidate in the top tenth.

The two sittings differed

1,650 candidates sat in January 2026 and 976 in October 2025. January's Critical Thinking median was 4.0 against October's 4.5; in Problem Solving it was 4.8 against 4.5. The report does not say why.

Time ran short in Problem Solving

217 candidates did not reach at least one Problem Solving question, and 742 of 2,626 (28%) either left one unreached or answered one in 5 seconds or less. In Critical Thinking it was 116 (4%).

Oxford's first cohort will change the mix

For 2027 entry, 10 Oxford courses require TARA, alongside courses at UCL. A new group of applicants sits the October 2026 test, so these figures are a benchmark from the previous cohort, not a forecast of next year's percentiles.

What this does not show

  • No university has published a TARA score cut-off, and no published data links TARA scores to offers. Nothing here is an admission chance.
  • The two modules are on separate scales and cannot be added or averaged into an official total.
  • These are 2026-entry candidates. Oxford's first use of TARA for 2027 entry changes who sits the test, so next year's percentiles may differ.
  • Only the printed quantiles are drawn. A percentile between them, as our free tool gives, is an estimate.
  • The older TSA was a different test with different candidates; its results are not comparable.

Cite this

TARAGenius (2026). What is a good TARA score? The first published benchmark. Published 27 September 2026, version 2026-09-27.1. https://taragenius.com/data/tara-first-score-benchmark

Chart credit
Source: TARAGenius analysis of UAT-UK, University of Oxford and others data
Version
2026-09-27.1, published 27 September 2026

You are welcome to quote the findings and reproduce the chart with the credit above and a link to this page. Please cite the original publishers too; their figures remain theirs. Corrections and new releases are added as new versions, and earlier numbers stay listed in the change log below.

Method and sources

  1. Transcribed Table 4 (page 9) of UAT-UK's TARA Technical Report 2025-26: candidates, mean, standard deviation and the 25th, 50th, 75th and 90th percentiles for each module and sitting.
  2. Took unreached and low-time counts from Tables 18 to 21 (pages 31 to 33). The report's table title defines low time as 5 seconds or less; its text says below 5 seconds.
  3. Counted courses listing TARA for 2027 entry from Oxford's and UCL's admissions pages.
  4. Drew only published figures. No distribution is smoothed or filled in between printed points.

Sources

Data and versions

  • TARA 2026-entry score distribution, scaling, reliability and timing: version 2026-09-26.1, checked 26 September 2026. First published version.
  • TARA course requirements for 2027 entry: version 2026-09-26.1, checked 26 September 2026. First published version.
  • This story: version 2026-09-27.1, first published 27 September 2026.

Download the chart data as CSV (UTF-8, one row per chart value, no personal data).

Keep going

© 2026 TARAGenius. An independent study resource, not affiliated with universities or test providers.Privacy