Observed 36 families with 1 18-24 mo old child for 7 hrs in their own homes with both parents and the child present. Ss included 18 boys and 18 girls. The parents interacted less in the 1st 2 sessions than in the last 5 and were more negative and less positive when they did interact in the early sessions. There was also a sex of child by time interaction, with boys receiving many more negative interactions early in the observations. It is suggested that observation data should be checked for stability prior to interpretation and that care should be taken to collect enough data to ensure stability.