• 65.1Last EU first-round place (Cagliari)
• 74.9Milano Statale, highest snapshot minimum
• 51.5Median EU score (41.1 in 2025)
• 5.5×EU candidates at 70+ (754 vs 138)

1. The EU cut-off jumped 10.5 points in one year. The last first-round EU place in medicine and dentistry went to 65.1 in 2026, against 54.6 in 2025.
2. Comparable snapshot minima rose 7.6 to 11.6 points. Milano Statale’s 2026 minimum was 74.9, Sapienza Medicine 72.7, Bologna 70.8, Pavia and Torino 70.3.
3. A 65 occupied a much lower EU rank in 2026. 65 points was EU rank 316–323 in 2025 and rank 1,495–1,512 in 2026, below the lowest Assigned/Booked total in this 2026 snapshot.
4. Every mark counts near the cut-off. There are 204 EU candidates between 64.5 and 65.5 inclusive in 2026. A few marks can separate many ranks, but there is no fixed rank gain per point.
5. Non-EU candidates are slightly stronger at the top. The conditional residual non-EU group averaged 52.2 vs EU 50.9 (scores 20+), and are the majority in every band from 65 to 84.9. The groups still overlap heavily.
6. Chemistry is strongly associated with total score. The site-and-total model estimates a non-EU chemistry advantage of about 1.2 points, but feasible ranges allow either group to lead.
7. Attempting and skipping need calibrated judgment. Under +1.5/−0.4 scoring, answering has positive expected value above a 21.05% chance of being correct. Accuracy on attempted questions does not establish accuracy on skipped questions.

In the supplied 2026 snapshot, the lowest medicine/dentistry Assigned or Booked total was 65.1 (Cagliari), and the highest course minimum was 74.9 (Milano Statale). The comparable 2025 minima shown below were 7.6–11.6 points lower. These are historical allocation observations, not final enrollments or future guarantees.

Source: supplied EU ranking snapshots, including Assigned and Booked candidates. The supplied 2025 PDF, page 47, shows Politecnica at 57.8 and rank 828; that source value is used consistently here. No comparable 2025 Catania minimum is shown. Firenze and Padova MedTech have no corresponding Assigned/Booked groups in this 2025 extract.

find your course ↓

• Higher snapshot marks at deeper ranks. Several destination groups reached deeper into the 2026 ranking, including 360 ranks deeper at Messina, while their minimum totals rose. Assigned/Booked counts do not by themselves establish official capacity growth or final enrollments.
• The cut-off is a crowded boundary. At exactly 65.1, 40 EU candidates were tied: seven were Booked, three were Waiting and 30 were Not Ranking. The same total did not guarantee the same outcome.
• These are publication snapshots. The 2025 list is headed 13 October; the 2026 list was captured on 26 October. Later allocation rounds can change course minima. These files do not establish final closing scores or guarantee a decrease for every course.

The recorded EU cohort at 20+ was 28.03% larger in 2026. Its median rose from 41.1 to 51.5. Neither list includes EU participants below 20, so this is not growth in all EU test-takers, and the score shift does not by itself establish a change in paper difficulty.

Source: 2025 EU ranking (7,202 candidates) and 2026 EU ranking (9,221 candidates), all scoring 20+.

The same rank, a different score

There is no single conversion between years. Around ranks 500–1,500, the observed score increase was approximately 11–13 points. There were 204 candidates in the inclusive 64.5–65.5 interval in 2026, versus 67 in 2025. A correct answer replacing a blank adds 1.5 points; replacing a wrong answer adds 1.9. Rank changes depend on the starting score and ties.

Among the 17,332 combined records at 20+, subtracting the supplied EU frequency at each total leaves 8,111 residual records. If the files cover the same population and the EU list is complete at 20+, this is the non-EU comparison group. Its mean is 1.33 points higher. These are score comparisons, not nationality classifications or admission probabilities.

Source: combined score frequencies minus EU score frequencies at each total of 20+. Residual counts are arithmetic results conditional on matched file populations and a complete EU list at 20+. No individual barcode is classified by total alone.

• The filtered distributions overlap heavily. With ties counted as half, the EU score-comparison probability is 46.92% for random pairs from these 20+ groups. This is not an admission probability or a comparison of all test-takers.
• Use complete score-threshold groups. At 70+, there are 754 EU and 921 residual non-EU results. An exact top-N boundary can cut through ties: totals alone allow 437–461 EU candidates in the first 1,000 positions, rather than identifying one exact split.
• Below 20 cannot be split. 1,036 candidates (5.6%) scored under 20, and the rankings don't show which group they belong to.

Higher-scoring groups commonly combine strong biology with higher chemistry and physics/mathematics totals. Chemistry has the largest section-total correlation (about 0.84), but the total already contains that section. This is association, not a measured payoff from extra study time.

All candidates scoring 20+ (17,332).

Right, wrong and blank answers

With +1.5 for a correct answer, −0.4 for a wrong answer and fixed section lengths, 91,633 of 91,840 section records have a unique integer correct/wrong/blank combination. The remaining 207 are ambiguous biology records; the fewest-attempts solution is used only as a convention. For example, 10.4 in chemistry uniquely corresponds to eight correct, four wrong and three blank.

Source: combined section records. Group estimates use source-location-plus-total buckets. Under the prefix mapping, 6,641 EU records occur in pure-EU buckets, identifying their aggregate section contribution; mixed buckets require proportional allocation. One Torino record at 69.6 remains unresolved. This does not link every individual EU test ID to a barcode.

• Attempted-question accuracy describes selected questions: model estimates are about 88% in biology and 65–66% in chemistry and physics/mathematics. They do not establish the probability of answering a skipped question correctly.
• The model estimates a chemistry difference. Estimated non-EU chemistry totals are about 1.21 points higher, alongside fewer estimated blanks. The data do not establish a psychological explanation or a causal effect of attempting more questions.
• The model estimates a reading difference: approximately 3.42 EU versus 3.03 residual non-EU out of 6. Feasible allocation ranges permit either group to lead; this is not a confirmed group advantage.

How uncertain are the group section comparisons?

The proportional site-and-total model produces point estimates, not uniquely identified group means. The conservatively calculated conditional non-EU-minus-EU sensitivity ranges all cross zero. Different feasible assignments within mixed buckets can reverse each apparent advantage.

Ranges reflect source-location-plus-total assignments; they are not probability intervals. The ranges allow the unmatched Torino record any observed section value at total 69.6 and are rounded outwards. They are valid outer ranges, not simultaneous attainable endpoints across sections. Without the location assumption, total-only bounds are wider.

What each score band looks like

Exact combined-group means, EU and non-EU together. These are average net scores, not individual requirements.

From the 60–64.9 group to the 75–79.9 group, the mean total rises about 14.79 points: biology rises 2.70, chemistry 4.14 and physics/mathematics 5.47. These are differences between score-selected groups, not causal study gains or a required individual profile.

The combined file labels 9,660 of 18,368 records Outside Italy. The EU site groups with comparable data have higher recorded 20+ means in 2026. Site labels do not establish nationality, admission destination or a causal regional effect.

Among the below-boundary records with full lists, Bologna, Milano Statale and Sapienza Medicine together account for about 40% of first-listed choices. The extracted lists average about 11 courses. Padova’s first-listed share was 7.6% in 2025 and 4.2% in 2026, while MedTech accounted for 3.9% in 2026. These aggregate shares do not prove which applicants switched. Site counts use all EU-prefix records, including the unresolved Torino record.

Use both years as historical benchmarks, not a forecast of 2027 cut-offs. Future marks and allocations depend on the paper, applicant pool, program mix, places and preferences. No guaranteed safe score follows from these two filtered cohorts.

1Use historical marks and ranks together

2Read the profile of the 70–74.9 group

• Biology32.4 / 34.5
• Chemistry16.4 / 22.5
• Physics & Maths13.6 / 19.5
• Logic5.7 / 7.5
• Reading & GK4.2 / 6
• ≈72.2

Combined-group average net section scores, not required correct-answer counts. The mean total is 72.24 before rounding; rounded components need not sum exactly. Strong biology is common, but a 72.9 total is observed with biology 25.0/34.5. Use your own error patterns to choose study priorities.

3Use calibrated probabilities when deciding to answer

read this before you guess

• With one equally likely correct answer, a blind five-option guess has expected value −0.02 points. If you reliably eliminate one wrong option and the four remaining options are equally likely, expected value is +0.075 (about +0.08); eliminating two gives about +0.23. Incorrect elimination can reverse the benefit.
• Expected gain from answering is 1.9p − 0.4, where p is your probability of being correct. It is positive above approximately 21.05%. Accuracy on attempted questions is not evidence that the same accuracy was possible on skipped questions.
• Use timed practice to check how reliably you can eliminate options. Classify mistakes by content, calculation and uncertainty; the cohort averages do not prescribe one attempt rate for every student.

4Build your course list carefully

• List courses in your real order of preference and follow the current official eligibility, allocation and enrollment rules. Historical score minima alone do not determine an individual allocation.
• Consider the full range of courses you would genuinely attend. In this snapshot, Cagliari, Messina, Catania, Siena Dentistry and Bari had minima of 65.1–66.5, compared with 74.9 at Milano Statale. Later closing scores and future outcomes are not established here.