This manuscript reports on the results of 2 experiments dealing with behavioral and adrenocortical responses to separation among 9-month-old human infants. In both experiments the social context of separation was manipulated. The results of Experiment 1 yielded evidence of a statistically significant adrenocortical response to 30 min of separation under conditions in which the substitute caregiver responded sensitivity to infant distress, but was busy and relatively noninteractive when babies were not distressed during the separation period. Altering the behavior of the substitute caregiver such that she was warm, responsive, and interactive throughout the separation produced a significant reduction in adrenocortical activity and in negative affect. In fact, these measures were not significantly different than those obtained when the mother and infant remained together in the playroom (No Separation). In Experiment 2, the effects of group versus singleton care were examined using the less stressful mode of substitute caregiving as described above. No significant condition differences in behavioral distress or cortisol were found. Furthermore, neither condition elicited a significant increase in cortisol over basal levels. Finally, these data provide evidence that maternal reports of infant Distress to Limits temperament, using Rothbart's Infant Behavior Questionnaire, predict adrenocortical responses to separation, while reports of Fear of Novelty do not.