For this study, nine mental factors were selected from Guilford's structure of intellect model, and eight interest factors identified from a factor analysis of the Thinking Interest Survey. These factors were then treated as dependent variables in the assessment of two methods of teaching algebra to 244 college students. Students were assigned to lecture instruction (n = 103) or programed instruction (n = 121) treatments within a regular mathematics class. Near the beginning of the course all students were administered mental and interest tests. Scores on achievement tests administered during the course were used to classify students in success and nonsuccess groups. A stepwise discriminant analysis was used to test research hypotheses. Results indicated that the 17 variables did discriminate among the success and nonsuccess groups for each instructional method. The set of combined variables which contributed to success in programed instruction were interest in convergent thinking, interest in divergent thinking, interest in nonsequential thinking, and convergent production of symbolic systems. For lecture instruction the set of combined variables contributing most to success included all of these as well as interest in non-specific thinking and convergent production of symbolic relations, the only variables which differentiated between success groups for the two instructional methods. (SD)