Twenty subjects were audio and video taped during their exercise leadership examination. The tapes were used to code the interactions that occurred between the exercise leaders and the participants in the simulated cardiac rehabilitation exercise session using an interaction analysis system developed specifically for this physical activity setting group. The system identified interactions that could occur during the warm-up, stimulus, and cool-down phases of the session. The tapes were coded using the Datamyte 801 Observational Recorder. The phases of the sessions were coded individually. A frequency count was made as the interaction categories occurred. The frequencies were converted into rates of interaction (f•min⁻¹) for comparison. The mean rates of interaction were low for the phases and overall (warm-up= .38/min; stimulus= .59/min; cool-down= .29/min; total= .46/min). The individual subject's rates of interaction were all less than 1 interaction per minute (minimum= .28/min; maximum= .72/min). Related t-tests across category facets between phases showed the instruction and explanation facets in the warm-up phase differed significantly from the same facets in the stimulus phase. The compliance facet differed significantly in the stimulus phase from the compliance facet in the other two phases. The monitor facet in the stimulus phase differed significantly from the monitor facet in the cool-down phase. There was no significant differences across facets between the warm-up and cool-down phases. Higher rates of interaction occurred more frequently in the stimulus phase. The coding showed the differences in the interactions of the exercise leaders in the different phases in the simulated exercise session. The low rates of interaction suggest that the exercise leaders may have been reactive to the specific examination situation in which these data were collected. / M.S.