As part of a longitudinal study, children in three groups (healthy, n = 33; cystic fibrosis (CF) n = 28; congenital heart disease (CHD) n = 23) were seen between the ages of 12 and 18 months to assess infant-mother attachment and at two years to observe parent-child negotiation of autonomy/dependence in a puzzle task. Although the healthy group included the highest proportion of securely attached infants and the CHD group the lowest, the distribution of attachment patterns did not differ significantly between groups, or between any of the groups and established norms. At two years the healthy children had the most positive experience in the puzzle task while the children in the CF group had the least positive experience. When diagnostic groups were divided according to prior attachment status, the data for the healthy group were consistent with the prediction that securely attached children would have the most positive mother-child interaction at age two. This was not the case in either of the medically diagnosed groups. These data suggest that the influence of the early parent-child relationship may be altered by a child's health status. We may need different models to account for development in healthy and medically compromised children.