Research findings on two aspects of the Law School Admission (LSAT) test are summarized. The two aspects are: 1) the selection of item types that would most accurately predict law school grades, and 2) the feasibility of shortening the test from five hours (plus one hour experimental) to 2 and 3/4 hours (plus 3/4 hour experimental) without appreciable loss of accuracy. A table summarizing relative predictive value of different item types is presented. Conclusions include: 1) the item-types with highest predictive value are (in order) reading comprehension; quantitative reasoning, including data interpretation and mathematics; and legal reasoning; 2) the most effective method of item-selection for the short test was choosing those items passed most often by students doing well on other items of the same type and failed most often by those doing poorly on other items of the same type; 3) on the shortened test, there must be as little overlap as possible in the abilities measured by each item type. Therefore, the amount of time given to each item type depends on the largeness of its relation to grades and the smallness of its overlap with other item types; and 4) the test could be (and was) shortened to 3 and 1/2 hours.