To test the efficacy of variables found effective in regular classrooms (in previous process-product research), variables were observed for 126 elementary school children in 17 resource classrooms. Measurement of teacher structure and student achievement was performed. Results indicated that, although most of the variables were used to at least a moderate degree in most classrooms, only the frequency with which students responded correctly consistently predicted performance on reading achievement measures. In contrast to regular settings, in which it has been suggested that a level of success that is too high may decrease student motivation, the study indicated that the higher the rate of correct answers the better the results in terms of achievement. Findings provided a caution in the generalization of process-product research to special education. (Author/CL)