Two hundred and thirty-six diabetic patients of Mexican-American origin were studied for the presence of pancreatic islet cell antibodies (ICA) and thyroid microsomal antibodies (TMA). One hundred and thirty-one subjects were insulin-dependent diabetic patients (IDD; type I), and 105 were noninsulin-dependent diabetic patients (NIDD; type II). The control population consisted of 79 normal subjects with no family history of diabetes mellitus.In addition, residual B-cell function (C-peptide) and C-peptide response to the iv administration of 1 mg glucagon were studied in IDD patients positive and neagtive for ICA.The prevalence of ICA was 25.9% in IDD patients, significantly higher than the 8.5% observed in the NIDD patients and the control population (P < 0.01). In IDD patients, the prevalence of ICA was strongly dependent on the duration of the disease and was 50% during the first year after diagnosis, 20.7% 2-5 yr after diagnosis, 11.7% 11-20 yr after diagnosis, and 2.9% over 21 yr after diagnosis. The prevalence of positive TMA in IDD patients was not significantly different from that in NIDD patients (4.5% vs. 2.8%; P = NS). Basal serum C-peptide and Cpeptide response to glucagon in IDD patients were similar in patients with and without ICA. Although there was a trend toward higher levels of basal C-peptide in NIDD patients positive for ICA compared to patients negative for ICA, the difference did not reach statistical significance (1.07 ± 0.36 vs. 0.62 ± 0.19 pmol/ml, respectively; P = NS).These findings, suggesting a lower prevalence of ICA and TMA in IDD patients of Mexican-American origin compared to Caucasian IDD patient populations in Europe and the United States, may indicate ethnic differences and the possible genetic heterogeneity of IDD populations. Residual B-cell function in IDD patients did not correlate with the presence or absence of ICA. However, in NIDD patients, C-peptide levels were found to be significantly increased in patients with or without ICA. This latter finding may be explained in part by the known hyperinsulinemic state in NIDD patients and may not be related directly to ICA. (J Clin Endocrinol Metab 54: 949, 1982)