Two improvements in supervisory practice would improve the effectiveness of student teaching: development of a set of objectives derived from a theoretical model of instruction to allow for meaningful observation, recording, and reporting of student teacher performance, and provision of feedback to inform the student teacher of his performance in implementing the model. A study was conducted to determine whether, in a situation where a comprehensive model of instruction had already been developed, systematic feedback would improve the application of the model through reducing discrepancies between what a teacher believes she teaches and what is observed as being taught. Eight student teachers were randomly assigned to experimental and control groups. Each was observed eight times over a five-week period. Subjects gave a Lesson Plan to the observer prior to observation, gave him a Reconstructed Lesson after the observation, and later received the observer's tally of discrepancies between the Reconstructed Lesson and his Observation Record. The experimental group also received immediate feedback during observation. Data was treated via repeated measures and linear trend analysis. The major hypothesis, that there would be significant differences in the overall trend of discrepancy scores as a result of feedback, was supported. The hypothesis that there would be significant differences in discrepancy scores favoring the experimental group was not supported. (Findings relative to other hypotheses are reported and implications discussed.) (JS)