This study investigated aptitude-treatment interaction (ATI) effects of variations in direct instruction. Two teachers each taught a 2-week fractions unit to two classes of fourth- and fifth-grade students. Each teacher taught one class using direct instruction and the other using a small-group variation of direct instruction. Students completed aptitude measures before the study began and achievement and attitude measures at the end. Regression analysis on achievement showed significant ATI and significant teacher effects. Students who had positive attitudes and an internal locus of control did better in the small-group variation, probably because they had some choice of activities and control over their learning. Teachers differed in effectiveness within direct instruction but did not differ within the small-group variation. (Authors)