Twenty-one patients with evident lipoatrophy treated with conventional (Conv.) insulin were either allocated to continuation of treatment with previously used insulin (Conv. group, n = 10) or were transferred to Lente MC (monocomponent) insulin with or without supplementary Actrapid MC insulin (MC group, n = 11). On entry and after 3, 6 and 12 months of follow-up, serum insulin-, pancreatic polypeptide- and proinsulin-binding IgGs were determined by radioimmunoelectrophoresis according to the method of Christiansen. Prior to determination of proinsulin-binding IgG, the insulin-binding IgG was removed by means of sepharose-bound insulin according to the method of Heding. In both groups a slight decrease in the titer of insulin-binding IgG was observed: in the Conv. group from 5.33 +/- 0.92 (SEM) to 4.66 +/- 1.17 mU/ml after 12 months, and in the MC group from 3.22 +/- 0.64 to 2.66 +/- 0.46 mU/ml, respectively. Due to the small number of patients with pancreatic polypeptide antibody titers above the detection limit no statistical evaluation was carried out. The level of serum proinsulin-binding IgG decreased in the MC group only (from 9.3 +/- 2.2 to 1.9 +/- 0.6 ng/ml after 12 months), and even showed a slight increase in the Conv. group (the respective titers were: 14.0 +/- 4.6 and 14.9 +/- 4.6 ng/ml). In the MC group 10 patients (91%) showed improvement and 7 (64%) complete regression of their lipoatrophy corresponding to 6 (60%) and 2 (20%) in the Conv. group. This finding suggests a possible role of proinsulin-binding antibodies in the pathogenesis of insulin lipoatrophy.