Fifteen Type I diabetic patients self-monitored daily mean blood glucose levels (BGL), psychological stress, diet, exercise, and insulin injections for a period of 8 weeks. Separate multiple-regression analyses were conducted for each subject to assess the predictability of BGL from stress reports, independent of the effects of diet, exercise, and insulin self-administration. Considerable individual variability in findings was evident, with stress predicting BGL in a statistically significant manner in 7 of the 15 subjects. Stress had a significant effect on BGL independent of diet, exercise, and insulin administration for each of these 7 subjects. Discussion focused on the problems of measuring the key self-management behaviors influencing BGL and on the possibility that a subgroup of diabetic patients may be BGL stress responders.