Individualized instruction programs have imposed an increased reliance on tests as a means of selecting and routing students through sometimes complex programs. Testing which occurs within the training sequence is particularly vulnerable to inefficient use of both trainee and instructor time, but computer-based instruction system can provide a means for monitoring and scheduling the routings. Both branched tests and short linear tests have been used, but both have disadvantages not present with the use of the sequential testing model. The sequential test may be described as one in which an examinee is given a test item, the response to the current item as well as previously administered items is evaluated in a certain way, and the examinee is either assigned to one of a number of classes and testing stopped, or judgment is suspended and testing continues. If single items from a test are employed for the derivation of discriminant functions, then functions could be computed to provide classifications after a small number of test items had been administered. This technique results in probabilities of group membership every time a linear combination is computed, and may be adapted to sequential testing by the addition of a termination rule. (BW)