Three questions concerning the Binary Search Method (BSM) of computerized adaptive testing were studied: (1) whether it provided a reliable and valid estimation of examinee ability; (2) its effect on examinee attitudes toward computerized adaptive testing and conventional paper-and-pencil testing; and (3) the relationship between item response time and item difficulty of the observations provided by the selected items. This research was conducted on a Commodore 64 microcomputer using the BASIC language. A paper-and-pencil test, the Pontiac Schools Student Needs Assessment Program-Mathematics Test, Grade 5 (SNAP5), was also administered to the 310 subjects, to obtain a valid estimate of examinee ability. This was a 44-item criterion referenced test. The BSM presented items until the last correct item was one less in order value than the lowest incorrect item. Results indicated similar BSM and SNAP5 scores for those students who took the BSM test first. Those taking the SNAP5 first had higher scores, a result attributed to the practice effect. Subsequent analyses indicated no significant differences between groups; correlations between BSM and SNAP5 were .69 and .68. There were no significant relationships between mathematics attitude or total response time and either ability measure. (GDC)