The interrelationship of life stress, social support, and glucose regulation was studied in eighty patients with diabetes mellitus (insulin-dependent = 40, non-insulin-dependent = 40). Glucose control was measured using glycosylated hemoglobin (HbA1); stress and social support were determined by the Social Readjustment Rating Scale and a Visual Analog Scale of social support. A two-way analysis of variance [social support (high/low) by life stress (high/low)] revealed that neither social support nor life stress was independently associated with HbA1. However, a significant interaction between these parameters was found. When reported stress was low, the HbA1 means were not statistically different for both high and low social support groups. As stress increased, variations in social support were associated with differences in glucose control. Under conditions of high stress, low social support subjects had significantly higher HbA1 than subjects with high social support (mean = 11.8% vs. mean = 9.9%, p = .04). These data suggest that during stressful times social support may insulate patients with diabetes from the adverse physiologic and behavioral consequences of stress and thereby foster better glucose control.