How to Get a 22+ on the DAT PAT Section

To get 22+ PAT, you need roughly 80 to 85% accuracy averaged across all six subtypes, not near-perfect accuracy on two subtypes and mediocre accuracy on the other four. PAT is 90 questions split into six subsections of 15, and it scores as one composite. That means a subtype sitting at 40% accuracy will cap your whole score no matter how good you are at the other five.

We scored a 25 AA (30 OC) and a 27 AA (29 TS) and got there systematically, not by luck. The PAT math is one of the most under-discussed parts of DAT prep, so let's break it down properly.

One thing underlies everything below: PAT is a trainable visual skill, not a knowledge test, and skills are built by reps. Every fix in this guide (finding your weak subtype, drilling it, retesting to confirm the gain) only works if you have a real supply of new questions to run it on. Run out of new material and you stop solving PAT and start remembering it, and your accuracy numbers go up for the wrong reason.

Why "good at PAT" isn't good enough for a 22+

Here's the trap almost every student falls into. You practice PAT, you notice you're crushing keyholes and cube counting, so you feel good about PAT and move on. Meanwhile pattern folding and angle ranking are quietly sitting at 50-60% accuracy, and you never really confront it because your "overall PAT feel" seems fine.

PAT doesn't work on overall feel. It works on six independent 15-question buckets that get averaged into one score. If you want 22+, every bucket has to pull close to its weight.

Run the numbers. Say you're at 100% on keyholes and cube counting (30/30), but only 53% on angle ranking, hole punching, and pattern folding (24/45), and a decent 73% on TFE (11/15). That's 65 correct out of 90, about 72% overall. That's a solid score, but it's well short of 22+ territory, and it's entirely because three subtypes are dragging the average down. Two perfect subtypes cannot rescue three weak ones.

The six PAT subtypes and what 22+ actually requires

The PAT covers six subsections, 15 questions each: keyhole/apertures, top-front-end (TFE), angle ranking, hole punching, cube counting, and pattern folding. To land around 22+ on the old 1-30 scale (or the equivalent zone on the newer 200-600 scale introduced in March 2025), you generally need to be missing no more than 2-3 questions per subtype, not acing three and bombing three.

These numbers are approximate: conversions have shifted over time, and the ADA's official concordance is the only source for exact equivalents. But the underlying math holds: consistency across subtypes beats brilliance in a couple of them.

Notice there's no column where one subtype gets to coast at 50%. That's the whole point. If you want an exact sense of what separates a 20 from a 22 in real percentile terms, we go deeper in our PAT score percentile breakdown of 20 vs. 22.

How to get 22+ PAT: the study plan

Once you accept the math, the plan writes itself. You're not "studying PAT" as one blob: you're running six separate mini-projects, each with its own accuracy target, and full-length practice tests are how you track all six at once.

1. Take a full-length practice test that mirrors real DAT format and timing. Not a random subtype drill: a real 90-question, 60-minute PAT under exam conditions. This is the only way to get an honest per-subtype accuracy read, because timing pressure changes which subtypes you actually struggle with.
2. Score it by subtype, not just overall. Write down accuracy for each of the six sections separately. Most students only look at their composite PAT score and miss that one or two subtypes are the entire problem.
3. Identify your bottom one or two subtypes. Whichever subtypes sit more than 10-15 points below your best subtype are your leverage points. Fixing a 50% subtype to 80% moves your composite far more than pushing an already-strong 90% subtype to 95%.
4. Drill that subtype in isolation using a method, not repetition. Untimed reps of a subtype you're bad at without a systematic approach mostly just wastes time. Learn the specific method for that subtype: for example, our TFE method guide or angle ranking rule-of-thumb approach: then apply it under time pressure.
5. Retest with another full-length practice test. Confirm the fix held under real timing and against new questions, not just the ones you already memorized the pattern for.
6. Repeat the cycle on the next-weakest subtype. Once one subtype catches up, a different one usually becomes your new bottleneck. Keep cycling until all six sit in a similar band.

Find your weakest subtype fast, then stop guessing at it

The single biggest time-waster we see is students who "know" they're weak at PAT in general and respond by doing more PAT, undifferentiated. That's slow. The fast version is: one full-length test, six accuracy numbers, one clear answer about where to spend your next ten hours.

If keyholes are your weak point, that's usually a visualization problem, not a speed problem: see our keyhole PAT tips for visualizing fast. If it's pattern folding or cube counting, it's almost always a spatial-reasoning fundamentals gap that a method fixes faster than raw practice volume. Diagnose first, drill second.

What happens if you ignore the math and just "practice PAT more"

We see this constantly in forum threads: a student retakes practice test after practice test, their PAT score barely moves, and they conclude PAT is just "not improvable" for them. What's actually happening is they never isolate which subtype is broken, so the same weak 50% subtype resets their composite every time.

Full-length tests are the diagnostic tool; subtype-isolated drilling with a real method is the fix. You need both, or neither works.

Keep score context honest as you track progress

Since March 2025 the DAT reports on a 200-600 scale in 10-point increments, with roughly 400 sitting near the middle of the pack (47th–54th percentile per the ADA’s official table), replacing the old 1-30 scale where 17 was about average and 22+ was considered great. A lot of advice online (including plenty of what you'll read about "22+ PAT") still refers to the old scale, so keep both in mind and don't panic if a number doesn't match what you expect. For the exact raw-score-to-scaled conversion logic, our DAT score conversion chart walks through it, and the ADA's official concordance is the source of truth for precise equivalents.

Remember too that PAT is scored separately from your Academic Average: it doesn't factor into AA, but almost every dental school looks at it independently, so it's worth getting right on its own terms.

The rep problem: why a finite bank breaks the exact plan above

Go back and look at the study plan in this post. It only works if the accuracy numbers you're reading are real. You take a full-length test, you get a 53% on hole punching, you drill the method, you retest to confirm the fix held. That last step (the retest) is where a finite question bank quietly wrecks the whole loop. If you're pulling from the same few hundred or few thousand items every other PAT resource ships, by your eighth or tenth full-length test you're not solving hole punching questions anymore. You're recognizing them. Your score climbs, your confidence climbs faster, and neither one reflects a fixed skill gap. That's the worst version of this plan: doing all six steps correctly and still not knowing if you're actually better.

This isn't a knock on effort. It's math. PAT isn't knowledge you can look up once and keep. It's a visual skill, and visual skills are built by reps, the same way a jump shot or a golf swing is. The number of genuinely new questions you see is the entire game. A finite bank is a countdown timer on that process. Work through it and the plateau isn't a sign you've hit your ceiling; it's a sign you've hit the bottom of the bank.

We built a PAT generator specifically because of this problem, not a bank, a generator. Every item across all six subtypes (keyholes/apertures, TFE, angle ranking, hole punching, cube counting, pattern folding) is produced from a seed-and-difficulty pair, so the supply of new items is unlimited by construction; you won't see the same question twice unless you deliberately ask for it. Every item is checked by an independent solver that re-derives the answer from the underlying geometry before it ships, which is a level of answer-key rigor a hand-built bank can't mechanically guarantee. Difficulty is a dial, not a fixed setting, so you can push past what the real test throws at you (tighter angle gaps, deeper cube stacks) and walk into the real PAT feeling like it slowed down. And because everything is seeded, you can still regenerate an exact test if you want to re-attempt one you bombed; unlimited doesn't mean you lose reproducibility, it just means you're never forced to reuse material to keep practicing.

Obvious disclosure: we built DATPractice, so weigh that as you read. But the underlying claim is checkable no matter which resource you use: ask how many genuinely new questions you can pull per subtype before you start seeing repeats. A finite bank versus a generator is a category difference, and for a skill that's graded on reps, it's the difference that actually moves your score.