Twenty-four diabetic patients were treated with porcine protamine-insulin (NPH-insulin) containing 7\p=n-\13 mmol proinsulin per mol insulin and 27 diabetic patients were treated with porcine protamine-insulin (HP-insulin) containing 0.36 mmol proinsulin per mol insulin. 75 % of the patients treated with NPH-insulin and 15 % of the patients treated with HP\x=req-\ insulin formed detectable insulin-binding antibodies. The difference in the antibody titre in the two groups was significant. As a group, patients treated with HP-insulin did not have a significant rise in the plasma insulin-binding capacity when compared to pre-treatment values. When comparing patients with antibodies and patients without detectable antibodies no difference in the degree of regulation could be demonstrated between the two groups. Young patients with antibodies had a higher insulin requirement per kg per day than patients without detectable antibodies.Among patients in remission those without detectable antibodies had a longer remission period than those with antibodies. Apart from the difference in antibody formation and hence a different distribution in the groups compared, the patients treated with NPH-insulin and HP-insulin did not differ with regard to the degree of regulation, the insulin requirement or the duration of the remission period.Commonly available recrystallized insulin preparations contain 1-2 per cent proinsulin (Steiner et al. 1968; Chance 8c Ellis 1969). The molecular weight of proinsulin is about 150 per cent of that of insulin (Frank 8c Veros 1968), and while the insulins generally used differ from human insulin in only one amino