The relationship between empathic role-taking abilities and empathic behaviors observed in both naturalistic and experimental settings was examined in 14 children between the ages of 53 and 65 months. Children were observed at the McGill Daycare Centre twice a week for 8 weeks, totalling approximately 30 hours. Event sampling was used to record the frequency and duration of empathic behavioral interactions, defined as the presence (or absence) of a prosocial response to a peer's display of one of four categories of emotion (e.g., comforting responses, participation, etc., in response to peer's observed emotional state of happy, sad, angry, or afraid). Midway through the observation period, role-taking tasks were individually administered. These included measures of perceptual-cognitive and affective role-taking skills. At the end of the 8 week period an experimental task was administered in which children could privately donate from 0 to 10 pennies to an unknown child. Results provide information (1) describing the frequency and duration of various kinds of emotional behaviors and the responses to them in a naturalistic setting, and (2) intercorrelating the various cognitive, naturalistic, and experimental measures of empathy. (Author/SS)