Relationships between cognitive and affective learner characteristics or "aptitudes" and mathematics achievement for sixth-graders in three instructional treatments in one school district were examined. Aptitude-treatment interaction (ATI) analyses were performed to test for differential prediction of achievement outcomes among treatments, using each aptitude as a single predictor. Multiple regression analyses were also run using aptitude, treatment, and aptitude x treatment interaction variables as predictors. Initial results with single predictors suggest a fairly consistent direction of differences among treatment slopes. This study provides an example of the ATI approach within a specific school setting in an evaluation context. (Author/BW)