We postulated that human insulin of recombinant DNA origin would be a poor immunogen and might prove to be less immunogenic than purified pork insulin. Results are reported for 100 diabetic subjects not previously treated with insulin. Individuals completed the first 12 months of a clinical trial of human insulin of recombinant DNA origin. These patients are contrasted with 121 similar individuals who are taking part in a trial of purified pork insulin. Prior to therapy, species-specific binding of 125I human insulin and pork insulins and insulin bound to antibody were undetectable in all individuals. In patients treated with human insulin of recombinant DNA origin, binding of 125I human insulin increased to 10 +/- 1.2% at 12 months versus increases in binding of 125I pork insulin in pork insulin-treated patients to 12.6 +/- 1.4% (NS). Mean percentages of species-specific binding tended to reach a plateau in the human insulin-treated group but continued to increase in the pork insulin group (p less than 0.001). Median bound values were nil throughout in patients treated with human insulin, but increased to 52 mU/l in the pork insulin group with significantly less bound insulin seen in the former group at all visits (p less than 0.001). The percentage of individuals who remained antibody free at 12 months, as indicated by bound insulin, was 56% in the human insulin-treated patients and 40% in the patients treated with pork insulin (p less than 0.01).(ABSTRACT TRUNCATED AT 250 WORDS)