Insulin antibodies were determined in sera from 38 children diagnosed as having juvenile diabetes for a duration of 0.7–15.2 years (median = 4.9 years). 8 children were treated with purified porcine insulins from the beginning of their disease, 16 children with bovine insulin NPH alone, and 14 children with non-purified, of hom 9 were later transferred to purified insulins. Serum insulin antibodies were measured by non-specific and specific methods using beef (B) and pork (P) antigens as described by Welborne and Sebriakova, respectively. 12/38 children had insulin binding levels similar to those of normal children, irrespective of the type of insulin used. The concentration of antibodies using radiolabelled B or P insulins as antigens were strongly correlated, by both the non-specific (p < 0.01) and the specific (p < 0.01) methods. Children with better score for diabetic control had significantly lower levels of insulin antibodies against B (p < 0.05) and P (p < 0.05) than those with poor diabetic control. There was also a significant positive correlation between mean HbAlc concentration and both B and P mean insulin antibody concentration (p < 0.01). Finally, patients treated with purified porcine insulin had significantly lower levels of antibodies than patients with non-purified bovine insulin (p < 0.05).