Practice exams are the highest yield study activity in law school, and most students extract a fraction of that yield because they use them wrong in one of four predictable ways: untimed, unwritten, unreviewed, or too late. The exam is a performance, three hours of producing written analysis under pressure, and performances improve through rehearsal under performance conditions followed by honest review of the tape. Everything below is that sentence, operationalized.

The four failure modes

Untimed: working a practice exam at leisure trains a skill the real exam never tests. Time pressure is not an inconvenience layered on the exam; it is the exam, the thing that forces triage, and triage is what gets graded. Unwritten: reading a fact pattern and mentally noting the issues, or sketching a bullet outline of the answer, feels like practice and skips the graded skill entirely, which is producing organized prose at speed.

The gap between spotting an issue and writing its analysis is where exams are lost. Unreviewed: writing an answer and never comparing it against a model is rehearsal without tape review; the writing was the workout, but the review is where improvement actually happens.

And too late: saving practice exams for reading week means the feedback arrives with no time to act on it. The first practice exam belongs around week eight or nine, covering whatever the course has covered, precisely because it will go badly while badly is still fixable.

The protocol

Take it cold: real conditions, full time, whatever materials the actual exam allows and nothing more, answer written in full prose. Then grade yourself against the model answer, and grade like a grader rather than a reader: work through the model with a pen and build three lists. Issues the model analyzed that you never raised.

Rules you stated imprecisely or incompletely. Places where the model argued facts and you concluded, the conclusory spots, which for most 1Ls is the longest list. Score honestly; the person you are calibrating is the one who will take the real exam.

Then run the error log, which is the step that converts practice into improvement. One running document per course, three columns: the miss, the category, the fix. Missed the statute of frauds issue, category issue spotting, fix, add SOF trigger to the Contracts attack checklist.

Stated the minimum contacts test without purposeful availment, category rule precision, fix, redrill that block of the outline. The log does two jobs: it turns each exam into specific outline and checklist repairs, and across three or four exams it reveals your miss pattern, which tells you whether your problem is spotting, rules, or application, three different problems with three different fixes. The spotting fix specifically is covered in issue spotting, and the model answer behaviors you are grading against are cataloged in what A exams do differently.

Cadence: one practice exam per course by week nine, then roughly weekly per course through the final month, four to six full timed exams per course by exam day. Later in the season, re take an early exam without rereading the model first; if the same misses recur, the fixes did not take, and the log tells you where.

Which exams, in which order

Your professor's own past exams are worth more than everyone else's combined, because you are not being tested on Torts, you are being tested on your professor's Torts: their format, their pet doctrines, their weighting between depth and coverage, their taste in policy questions. Most schools archive past exams in the library database or the course page, and upper level students in your section's mentorship programs know where the rest live; take only from permitted sources, since exam banks are governed by honor codes. Work the professor's exams newest to oldest, and save the most recent one for a full dress rehearsal in the final two weeks.

When their model answers exist, they are the single most valuable documents of the semester. Before you burn those exams on practice, profile them: format, recurring doctrines, the never tested list, what the model answers reward. That profile is exactly what our Exam Analyzer builds from uploaded past exams and a syllabus, and it turns the professor's archive into a prep plan before you spend a single practice session.

When the professor's archive is thin, substitute in this order: same course, other professors at your school, then released exams from other schools, then supplement question banks, each a step further from the exam you will actually face, each still useful for reps. Course specific banks are linked from every course hub, starting with the 1L guide.

Grading yourself without a model answer

Some exams circulate without models. Grade structure instead: list every issue you addressed and ask, for each, whether the rule was complete and whether every application sentence contained a fact, the because test.

Swap answers with one trusted classmate and grade each other's against the same checklist; you will each spot the other's conclusory paragraphs faster than your own. This is one of the two legitimate jobs of a study group.

FAQ

How many practice exams should I take per course?

Four to six full timed exams per course, plus earlier partial ones, is the range where most students plateau. Below three, the exam format is still novel on test day; the review discipline matters more than the raw count.

Should I retake practice exams I have already done?

Yes, once, a few weeks later, as a fix verification pass. Familiar facts inflate the issue spotting, so grade yourself only on the previously logged misses: did the repairs hold.

What if my professor has no past exams available?

Profile their class instead, their hypos are their exam questions in rehearsal, use adjacent professors' exams for reps, and ask directly in office hours what the format will be. Professors answer that question far more often than students expect.

Written by Ali, Georgetown Law, founder of Lovare Institut.

Book a call