The leavetakings and reunions of infants, toddlers, preschoolers, and their parents were observed as 56 children were dropped off and picked up at nursery school. During the pilot phase of the study, shorthand running records were made of the behaviors of each child and his or her parent for all of their leavetaking and reunion episodes. From 100 running records, checklists were derived to represent behaviors characteristically observed. During the 6-week observation phase of the study undertaken at the beginning of the fall semester, 15 randomly selected leavetakings were observed for each child. Six months later, 10 leavetakings and 10 reunions were observed for each child during an additional 6-week period. A complete set of observations was produced for 40 children. In addition to multivariate and univariate analyses of variance and post hoc comparisons by Bonferroni "t" tests, regression analyses were conducted to determine which child and parent behaviors accounted for the variance in separation distress and which behaviors accounted for differences in behavior upon reunion. Results are discussed. (RH)