A study was conducted to assess the reliability of the College-Level Academic Skills Test (CLAST) for the major ethnic groups in Florida; and to develop a standard error of measurement for the critical group that fell close to the cut score on each test. The October 1983 and March 1984 administrations of the CLAST were analyzed to determine whether one ethnic group was consistently obtaining lower reliability estimates than other groups, the accuracy of each subtest in student placement, standard error, and subtest reliability. Study findings included the following: (1) the computation portion of the CLAST had the highest reliability, while the writing subtest had the lowest reliability; (2) the scores of Hispanics and Black non-Hispanics usually had the lowest reliabilities of any sub-group; (3) the size of the standard error of measurement around the cut score was somewhat higher than the traditional standard errors of measurement previously reported, indicating that students who fell below the cut score by 10-20 scale score points should be encouraged to retake the test. Information on previous reliability studies is included. (LAL)