Nineteen mothers and their developmentally delayed infants, participants in an infant stimulation program, were observed in their homes. Observers described the infants' behavior using the Waters and Deane (1985) Attachment Behavior Q-sort and the mothers' interactive behavior with the Pederson and Moran (Pederson et al., 1990) Maternal Behavior Q-sort. The infants' therapists completed ratings of maternal sensitivity and involvement using the Ainsworth (Ainsworth, Bell, & Stayton, 1971) sensitivity scales, the Caldwell and Bradley (1984) HOME inventory, and the Bromwich (1981) Parent Behavior Progression. Maternal sensitivity scores derived from the observers' Q-sorts correlated significantly with ratings of sensitivity provided by the infant therapists and with assessments of other aspects of the mothers' behavior reflecting the intellectual quality of interactions. A significant relation with security of attachment was found only with measures of maternal sensitivity, assessed either by the same observers using the Q-sort, or as judged independently by the therapists using the Ainsworth rating scale. Thus, the mother may play distinctive roles in her infant's emotional and intellectual development. Mothers of developmentally delayed children reported that their infants presented substantial parenting difficulties, but at the same time, self-reports of stress associated with parenting were not different from normative levels. This population may provide unique opportunities for the study of the relation of parenting stress to early interaction and social development.