This paper addresses problems confronting law school teachers in grading law school exams and assigning letter grades. Using prototypical dialogue and scenarios, the paper examines mathematical and statistical issues that contribute to grading errors. Discussed in relation to real world data and the bar exam are: differential weighting, combining scores, test reliability, consistency in measurement, and standard error issues. The paper also reviews two sets of court cases. In so-called "academic challenge" cases, case law is clear--the burden of proof is on test-takers who must show that tests violate accepted norms. In "high-stakes testing" on the other hand, the burden of proof is upon test-scorers, who must prove that tests comply with accepted academic norms. Since such cases often involve claims of civil rights, court rulings are more ambiguous. This raises the question of whether most law school grades are high-stakes tests or simply academic challenge situations. Appended to the paper are sample test questions which instructors can used to evaluate their own grading biases. Also appended is a chapter, "Constructing and Using Essay and Product Development Tests" from the book, "Measuring and Evaluating School Learning", by Lou M. Carey. (Contains 50 references.) (CH)