A model for computer-assisted branched testing was developed, implemented, and evaluated in the context of an elementary school using the system of Individually Prescribed Instruction. A computer was used to generate and present items and then score the student's constructed response. Using Wald's sequential probability ratio test, the computer determined whether the examinee was or was not proficient in the skill being tested. If such a decision could be made, he was branched to another objective according to specified criteria based upon the hierarchy. Otherwise, another item was generated and the cycle repeated. Results showed that the computer test was highly successful in providing reliable information in substantially less time than that which was required by the conventional paper and pencil test. (Author/SP)