The technique of path analysis was used to test the role of student task engagement as a mediating process variable linking teacher behavior and student achievement in mathematics and social studies classes. Study participants included 40 teacher interns and their respective second-, third-, and fifth-grade classes from a school district served by the University of South Florida (Tampa). Class size ranged from 20 to 30 students. Two measures of teacher behavior and a measure of student task engagement were obtained during lessons from 2-week instructional units designed specifically for the study. Achievement was defined as the class mean residualized gain score, residualizing student posttest scores on both content pretest performance and scores on the relevant subtests of the Comprehensive Tests of Basic Skills. Results suggest different causal paths from teacher behavior to achievement in the two contexts. In social studies classes, task engagement emerged as a mediating variable as hypothesized, while in mathematics classes, task engagement played no causal role in achievement. Results were interpreted as an example of the added information obtained from testing path models rather than examining only zero-order correlations. Four data tables and three figures are included. (Author/SLD)