Teachers sometimes fail to use previously acquired teaching skills. A review of studies which had examined the maintenance of teaching skills found that some training programmes which used skill practice and feedback on performance were successful in achieving maintenance. The present study was designed to test the effect of practice and feedback in diverse settings on the acquisition and maintenance of teaching skills. The following skills were selected for training: 1. Increasing the use of approval and decreasing the use of disapproval. 2. Increasing the use of feedback and decreasing the use of criticism. 3. Using wait-time: (a) after asking a question and before calling on a student to answer, (teacher wait-time) and (b) after a student response has finished (pupil wait-time). Repeated measures were made of nineteen student teachers teaching during a six week student teaching practice prior to the training course and again immediately following it. Ten of them, who also secured teaching positions, were observed when teaching in their own classrooms. During the training course, the subjects practised some skills until the training targets had been achieved five times in each of two settings (the 2 X 5 treatment). They practised the remaining skills until the training targets had been achieved two times in each of two classroom settings (the 2 X 2 treatment). The subjects observed one another practise and the results of these observations were used to provide them with performance feedback. There was a general training effect although there was no treatment effect for the amount of practice. More maintenance was found when the subjects became employed as classroom teachers than was observed immediately after training. More skills were maintained when there was a match between the class level being taught and the class level practised with during training. Feedback was maintained by most subjects while teacher wait-time and low rates of criticism were maintained by the fewest subjects. It was hypothesized that the subjects had previously been subjected to thousands of hours of observational learning of teacher behaviour and that a brief training course may not have had sufficient impact to counter such prior learning. It was also hypothesized that there were unidentified stimuli, context variables and sources of reinforcement controlling the performance of particular skills by individual subjects. It was concluded that future research in this field should seek to identify these sources of stimulus control.