The purpose of this study was to investigate the effects of three levels of training (direct, indirect, and none) on teachers' ability to accurately rate video of student behavior. Direct and indirect training groups received instructional sessions on direct behavior ratings (DBRs), with the direct training group receiving opportunities for practice and feedback. Each training group viewed a video of a typical classroom environment and then rated the proportion of time the student displayed disruptive behavior and academic engagement. Accuracy was examined through comparison of the teachers' ratings with direct observation data. Teacher acceptability of using DBRs was also assessed. Overall, the results suggested that the most intensive level of training (i.e., direct) did not improve the accuracy with which teachers rated behaviors using DBRs. Limitations and future implications are discussed. (Contains 3 tables and 1 figure.)