This longitudinal study investigated the relationship of stress arising from medical problems of newborns to parent-infant interaction through the infant's first year. Significant interactive differences between full term and premature infants were found in feeding situations during the neonatal period and in floor play at eight months. The sample of 40 parent-infant dyads was divided equally among full term, healthy preterm, and sick preterm infants, and infants of diabetic mothers. Each dyad was observed eight times (in hospital, home, and laboratory settings). On the hypothesis that medical problems would create stress leading to interactive difficulties, it was predicted that full term infants would have the fewest interactive problems and there would be increasing problems for healthy preterm infants, sick preterm infants, and infants of diabetic mothers, respectively. The Bayley Scales of Infant Development and the Brazelton Neonatal Assessment Scale were used to gauge infant behavior and interaction with parents and strangers, the experimenters. With the exception of the infants of diabetic mothers group, whose ordering was inconsistent, the relative position of the groups on each measure was consistent with the hypothesized continuum. In feeding situations, there were significant differences during the neonatal period which diminished at four months. In the 15-minute floor play sessions, there were significant group differences at eight months, diminishing at 12 months. The see-saw pattern of increasing and decreasing differences among the groups, and alternative interpretations of this pattern are discussed. It is suggested that the group differences indicate that it may be more difficult to care for preterm infants than full term infants. (Author/BH)