Fifty randomly selected moderate and low implementing Distar Language teachers were retrained in techniques of correcting students' mistakes and recycling through an instructional task until all of the children in the group responded without error (criterion teaching). As a result of retraining, the experimental teachers performed at a significantly higher level of implementation than the control teachers. In addition, there were significant differences (favoring the experimental group) in achievement (for both a program-specific and program-general measure) on the students' statistically adjusted posttest scores. Implications for research and training in teacher education are discussed. (Author)