Repeated emergency department (ED) visits by HIV-infected persons may signify poor access to care or treatment from inexperienced ambulatory providers. We examined features of 157 clinics following 6820 HIV-infected patients and associations with repeated (> or =2) ED visits by these patients in the year before their first AIDS diagnosis. Patient clinical and health care data came from 1987-1992 New York State (NYS) Medicaid files and clinic data came from interviews of clinic directors. The HIV/AIDS experience of each study patient's clinic was measured as the annual number of Medicaid enrollees newly diagnosed with AIDS who were contemporaneously followed by the patient's clinic. Repeated ED use was observed for 24%. The adjusted odds ratio (AOR) of repeated ED visits was reduced for patients in clinics with a physician on-call (0.77; 95% confidence interval [CI] = 0.65, 0.92), evening or weekend clinic hours (0.77; 95% CI = 0.64, 0.93), or >50 AIDS patients/year in 1987-1988 (0.56; 95% CI = 0.44, 0.71) versus fewer patients in those years. Patients in clinics with more than one feature promoting accessibility or HIV expertise had a greater reduction in their AOR of repeated ED use. HIV-infected patients in clinics with greater accessibility and HIV expertise rely less on the ED for care.