This study evaluated Krell's 1981-82 Scholastic Aptitude Test (SAT) preparatory series software purported to raise students' scores substantially after only a short term of computer-assisted instruction (CAI). Forty-eight college-bound juniors from Escambia County (Florida) were assigned to experimental and control groups. A two-phased pre- and post-test design was used. Both groups received 15 hours of CAI, but during different phases of the study. Data were collected through the administration of one SAT pretest, two SAT posttests, the Otis-Lennon School Ability Test, the Witkin Group Embedded Figures Test, and a student questionnaire for demographic information. Ten hypotheses concerning the effects of CAI, cognitive styles, socioeconomic status, gender, IQ, grade point average, parents' educational attainment, parents' occupational status, number of siblings, and college majors on SAT score gains were tested using non-parametric statistics. With one partial exception, all hypotheses were accepted in the null form. The results showed: (1) Krell SAT software utilized under controlled CAI conditions produces only modest group score gains, and (2) individual student score gains under these conditions, by contrast, may be high. That is, group SAT score gains will seldom exceed chance expectations and will be of little practical consequence. On the other hand, individual students may evidence significant gains, which is consistent with the publishers claim. The study includes a nine-page reference list. (JAZ)