About 60% of respondents were re-hospitalized at least once for diabetes-related reasons (n = 128). Half of those questioned had Medicaid or no health insurance at all; 23% fit criteria for a non-autoimmune, type 2 diabetic phenotype. Those who avoided re-hospitalization were more likely to have been seen initially at a tertiary care facility, to have private health insurance, and to be males. They had, on average, 2 yr shorter duration of diabetes at the time of interview. Risk for re-hospitalization was not associated with age at diagnosis, ethnicity, diabetic phenotype, or source of care during the past year. In multivariate analysis, predictors of re-hospitalization were gender [odds ratio (OR) 1.98 for females vs. males (95% confidence interval (CI) = 1.05-3.72)], duration of diabetes [OR = 1.46 per yr (95% CI = 1.36-1.57)], initial ascertainment at a community hospital [OR = 5.44 vs. tertiary care facility (95% CI = 2.61-11.29)] and having Medicaid or no insurance [OR = 2.73 (95% CI = 1.42-5.24)], compared with those with another type of health insurance. There is a high risk of re-hospitalization after the initial diagnosis of diabetes among insulin-treated minority children, particularly the uninsured and those on Medicaid, in part related to duration of disease and where the initial treatment occurred.