The number of hospitalized patients lacking an identifiable source of third-party payment has risen substantially in recent years. This study examines trends in the hospitalization of "self-pay" patients and investigates causal influences on the propensity of hospitals to accept such patients for treatment. Our analysis pays particular attention to the relationship between Medicare's prospective payment system (PPS) and hospitals' self-pay patient share. Our results show an overall increase in both the number and proportion of self-pay patients treated by hospitals between 1980 and 1985. Substantial differences existed among the types of hospitals that accepted such patients, with major teaching hospitals treating an increasingly disproportionate share. The mix of self-pay patients in terms of age, sex, and reason for hospitalization remained stable during the period under study. Our conclusion is that the regression analysis shows no evidence that PPS reduced hospitals' willingness to treat uninsured patients.