Clinical supervision is of growing importance professionally, but instruments to measure its effectiveness are scarce. Based on the observational instrument Teachers' PETS, two complementary levels of outcome measurement were used to analyse supervisory effectiveness, namely momentary time sampling (i.e. a micro-analysis of frequencies) and the more molar “change episodes”. Ten audio-taped sessions of routine (baseline; N = 5) and CBT supervision (N = 5; i.e. the intervention phase) were coded with both measures, to assess their relative sensitivity to this manipulation. Improved supervisee learning was detected during the intervention phase by both measures. However, a retrospective comparison between the data within these change episodes and the accompanying non-episode data indicated that the micro level of analysis provided a more sensitive measure of supervisory effectiveness. Technical and conceptual issues arise.