This analysis shows that persons in the United States without health insurance experience barriers to access to medical services, both ambulatory care and hospitalization. More important, for poor and near-poor persons, insurance status as a simple insurance- no-insurance dichotomy is an inadequate characterization of their access to medical services. Having Medicaid coverage, in contrast to having private insurance or no health insurance, is strongly related to increased access to such services. For low-income groups, private insurance provides no greater access to medical care than being without insurance entirely. The putative explanation is that, for low-income groups, the deductibles and copayments of private health insurance are themselves excessive barriers to care. Variations in state Medicaid programs' eligibility criteria affect likelihood of enrollment in Medicaid. There are no consistent results, however, to support the proposition that variations in the generosity of state Medicaid programs influence health-care access for persons on Medicaid. A statistical analysis was undertaken to quantify differences in access to medical services in relation to an individual's having health insurance. The analysis used three years of the National Health Interview Survey's Health Insurance Supplement (1983, 1984, and 1986) for persons less than 65 years of age and sample sizes ranging from 50,000 to 80,000, depending on year. Much of the analysis utilizes a two-stage procedure: Persons are first scored on their "vulnerability to care," taking into account demographic and health variables. Then, those scores are compared across quartiles. Data are presented in a manner that is accessible to readers not sophisticated in multivariate analysis without losing the control of multivariate analysis. The data analysis is preceded by a discussion of the conceptual and methodological hazards of research on access to medical services.