Volunteer college students were assigned to one of six computer administered vocabulary tests, one half with immediate knowledge of results (KR) after responding to each item, and the other half without knowledge of results. The six tests were designed to be at one of three levels of difficulty and consisted either of 50 preselected items (conventional testing) or tailored on the basis of previous candidate responses (stradaptive testing). Results indicated that the mean maximum-likelihood estimates of individuals' ability varied as a joint function of the KR condition and of test difficulty. Students in the KR condition scored higher than the other students on the most difficult test and lower on the least difficult test. Questionnaire results indicated that, although the students perceived the differences in test difficulty, there were not shown to be any effects on mean student anxiety or motivation scores attributable to test difficulty alone. Students in general reacted very favorably to receiving immediate knowledge of results and its provision increased the mean level of reported motivation. (Author/CTM)