Reproducibility of C-peptide secretion was assessed in 20 children (group 1) by their responses to two Sustacal- (a mixed liquid meal) stimulation tests performed 7-14 days apart. For the 12 C-peptide-positive children (basal C-peptide greater than or equal to 0.03 pmol/ml) there were no differences in the basal or stimulated values between tests 1 and 2. The effect of exogenous insulin on C-peptide secretion was assessed in 20 other children (group 2) by their responses to two Sustacal tests, one test without and one with soluble insulin (0.25 U/kg) injected subcutaneously before testing. Eleven children were C-peptide positive and had no differences in C-peptide response between tests 1 and 2. The results from test 1 in groups 1 and 2 were combined with those from 44 others undergoing a single Sustacal test (group 3, N = 84). There was a close correlation between basal and peak C-peptide concentrations in the 44 C-peptide-positive children (r = .88, P less than .001). Peak C-peptide concentrations correlated inversely with HbA1 (r = -.29, P less than .01), insulin dose in units per kilogram (r = -.40, P less than .001), and duration of diabetes (r = .33, P less than .001) and positively with age at onset of diabetes (r = .34, P less than .001). The C-peptide-positive children had reduced glucose response to Sustacal, lower HbA1 concentration, lower insulin requirement, later age of onset, and shorter duration of diabetes than children who were C-peptide negative.