How Law School Exams Are Graded

In nearly all doctrinal courses, the professor reads every exam personally. Law schools generally do not use teaching assistants to grade final exams in J.D. courses; TAs may grade practice problems or legal writing drafts, but summative grades in a torts or contracts class come from the professor. Large courses are the exception only in that some schools allow a professor to hire a former student to grade a multiple-choice section or to do a first-pass sort.

Grading is anonymous. You are assigned an exam number each semester, and the registrar holds the key until grades are submitted. The professor sees only the number. This is why you should never sign your name, reference a specific class discussion in an identifying way, or email a professor about your exam before grades post.

Realistically, a professor spends roughly fifteen to forty minutes on each exam, depending on length and format. In a 90-student section that is still 25 to 50 hours of reading. The practical consequence is that your reader is moving fast, is looking for specific things, and cannot excavate a buried good argument. Structure and signposting are not stylistic preferences; they are what makes your points findable.

Most professors build a grading sheet before or while reading the first several exams. The typical form is a list of legal issues embedded in the fact pattern, with points allocated to each: spotting the issue, stating the governing rule accurately, applying the rule to the specific facts, and addressing the strongest counterargument. Some professors assign explicit point values (e.g., 8 points for the promissory estoppel discussion, broken into 2/2/3/1). Others read holistically first, sort exams into rough piles, then re-read the borderline papers.

The single biggest source of point spread is application, not rule recitation. Nearly everyone in the class knows that a battery requires intent to cause a harmful or offensive contact. Points come from tying the specific facts to the specific element and explaining why a fact cuts one way rather than the other. 'Because he swung the bat, there was intent' earns almost nothing. Explaining why swinging at a spot two feet from the plaintiff's head satisfies or fails substantial certainty earns the point.

Professors also reward what the fact pattern was engineered to test: the ambiguous facts. If a fact is genuinely two-sided, the professor planted it. Arguing both sides and then taking a position typically scores higher than a confident one-sided answer, because the point sheet allocates credit to each side of the argument.

Points are almost never deducted for wrong answers on issue-spotters, but they are lost through opportunity cost. Time spent on a non-issue is time not spent earning points elsewhere, and a long discussion of an issue the facts foreclose signals to the reader that you cannot distinguish what matters.

After scoring, the professor rank-orders the exams from highest raw score to lowest. Only then does the curve apply. Almost every ABA-accredited law school imposes a mandatory curve on first-year required courses and on upper-level courses above an enrollment threshold, commonly 25 or 30 students. The curve typically specifies a required median or mean grade and often maximum and minimum percentages in each grade band.

The specific numbers vary substantially by school, and this variation is real, not cosmetic. Some schools set a mandatory B median; others set B+; a few highly ranked schools use pass/honors systems with no letter curve at all. Some schools mandate an exact distribution (e.g., 10-15% A range, 30-40% B+, etc.); others fix only the median and let the professor distribute around it. Some schools have a mandatory ceiling on A's and a discretionary floor on failing grades. Check your school's academic rules — the curve is published, usually in the student handbook or on the registrar's page.

The consequence is that your raw score is meaningless in isolation. A 78 out of 100 might be an A- if the class median raw score was 62 and a C+ if the median was 76. Small point gaps at the median translate into whole grade differences because that is where the exams are densest.

Small seminars, clinics, externships, and independent study are usually exempt from the curve, which is why upper-level GPAs tend to drift upward.

In most classes the A and B exams contain the same core issues. The differences cluster in a few places. A exams spot the secondary issues — the ones only a third of the class sees. A exams resolve conflicts between rules rather than reciting both and moving on. A exams handle the facts with precision, quoting or paraphrasing the operative detail rather than gesturing at 'the facts here.' And A exams allocate time proportionally, giving the biggest issues the most space.

Length correlates with grade but does not cause it. Faster typists produce more words and, on average, more points, because issue-spotters reward coverage. But padding, restatement of the facts, and long rule recitations without application add words without adding score. If you find yourself copying the fact pattern back, you are producing text your grader will skim past.

Policy discussion is class-specific. Some professors allocate real points to policy; many allocate none on an issue-spotter and reserve it for a separate essay question. Their old exams and model answers tell you which kind of professor you have.

Multiple choice is scored mechanically, then folded into the composite score before the curve is applied. Most law school multiple choice does not penalize guessing, but confirm this on the instruction page. Professors sometimes drop a question after the fact if item analysis shows the class performed randomly on it, which usually signals a defective question.

Short-answer and policy questions are graded against a model answer with a narrower band of acceptable responses. Papers and seminar work are graded holistically, often against a rubric covering thesis, research depth, use of authority, and writing quality; citation errors under The Bluebook are typically a small deduction in doctrinal seminars and a much larger one in legal writing courses.

Participation, where it counts, is usually applied after the curve and is capped — often a one-notch adjustment (B+ to A-) for a limited number of students. Many schools require the professor to disclose in the syllabus that participation will be used and how much it can move a grade. This is the one moment when anonymity is lifted.

Almost never, and the ground is narrow. Nearly every law school limits grade challenges to mechanical or clerical error: a missed page, a mis-added score, a transposed exam number, a grade entered for the wrong student. Disagreement with the professor's evaluation of the merits is not a reviewable ground.

Courts reinforce this. In Board of Curators of the University of Missouri v. Horowitz, 435 U.S. 78 (1978), and Regents of the University of Michigan v. Ewing, 474 U.S. 214 (1985), the Supreme Court held that federal courts owe substantial deference to genuinely academic judgments by faculty and will not override them absent a substantial departure from accepted academic norms. Practically, litigation over a grade goes nowhere.

What you should do instead is ask to review your exam. Most professors will meet with you, and many post a model answer or a memo explaining the common errors. Read your own exam against the model. The diagnostic value of seeing where you wrote a conclusion instead of an analysis is far higher than any appeal.

Given how grading works, the highest-leverage moves are structural. Answer in the order the call of the question asks. Use headings that name the issue or the party so the grader can find each discussion in two seconds. Write in short paragraphs with one issue each. State the rule in one or two sentences, then spend the bulk of the space on facts-to-elements application.

Take every practice exam the professor has made available and write out at least two under timed conditions. Professors reuse fact-pattern architecture and issue priorities year to year, and their model answers are the closest thing you will get to the point sheet.

Finally, calibrate to the class. Because the curve is comparative, the question is not 'did I write a good answer' but 'did I write a better answer than the person next to me.' That means finishing the exam matters enormously — an unwritten final issue is guaranteed zero points, while a rushed two-sentence treatment of it may still capture half the available credit.