Insulin-like growth factor-I (IGF-I) enhances insulin action in normal subjects and in patients with both type 1 and 2 diabetes; however, its administration is associated with significant side effects in a high percentage of patients. The coadministration of IGF binding protein-3 (IGFBP-3, the predominant IGF binding protein in serum) with IGF-I limits IGF-I inducible side effects, but it does not attenuate the ability of IGF-I to enhance protein synthesis and bone accretion; therefore, we determined whether IGF-I/IGFBP-3 would retain biological activity in type 1 DM and limit side effects associated with free IGF-I administration. Twelve patients received recombinant human IGF-I plus IGFBP-3 (2 mg/kg-day) by continuous sc infusion for 2 weeks. Each subject served as his own control; and, during a paired 2-week period, each received a placebo infusion. The order of the treatments was randomized. Subjects were placed on a constant caloric intake but were allowed to adjust insulin doses to maintain appropriate levels of glycemic control. Subjects measured blood glucose four times per day at home and kept a log of their insulin use. Frequent sampling for glucose, insulin, and GH was conducted during four inpatient study periods, one at the beginning and one at the end of each 2-week study interval. During IGF-I/IGFBP-3, insulin doses were reduced by 49%, and mean serum glucose was reduced by 23%. Free insulin levels obtained during frequent sampling in hospital fell 47% on IGF-I/IGFBP-3, compared with control, but showed no change with placebo. Concomitant glucose measurements did not differ in the two treatment groups. There was no change in body weight. Fructosamine levels decreased by 12%, but this was not significant (P < 0.1). Fasting triglyceride was unchanged, but cholesterol declined from 170 +/- 24 to 149 +/- 31 mg/dL (P < 0.05). IGFBP-2 (an IGF-I-dependent responsive variable) rose from 141 +/- 56 to 251 +/- 98 ng/mL (P < 0.01) on IGF-I/IGFBP-3. To analyze the mechanism by which IGF-I/IGFBP-3 might reduce insulin requirements, the change in serum GH was quantified. Mean GH levels were reduced by 72%, from 2.48 to 0.55 ng/mL (P < 0.001). An equal number (40%) of drug- and placebo-treated subjects had minor hypoglycemic episodes at home that required adjustment of insulin doses. No episode was classified as severe. In contrast to previous studies with free IGF-I, there were no cases of edema, headache, jaw pain, retinal edema, or Bell's palsy. No subject withdrew because of drug complications. These findings indicate that IGF-I/IGFBP-3 is biologically active on carbohydrate metabolism, as measured by a decrease in insulin requirements in patients with type 1 diabetes. Further studies will be required to determine the long-term safety and efficacy of this combination in patients with insulin resistance and diabetes.