PURPOSE: continues to devastate urban communities, particularly among marginally-housed, ethnic minority, and drug-using populations. This study (1) describes access to comprehensive medical care, quality of HIV-related care, and attitudes regarding health among HIV-infected residents of single-room occupancy (SRO) hotels and (2) explores predictors of the use of pneumocystis carinii pneumonia (PCP) prophylaxis and highly-active antiretroviral therapy (HAART). METHODS: We conducted a cross-sectional, community-based study of 69 Bronx SRO hotel residents during May 1998. Utilizing door-to-door recruitment, we administered a 41-item, anonymous questionnaire to assess participants' demographic characteristics, level of illness and health care utilization, use of HIV-related therapies, and perceptions of their own health and medical care. RESULTS: Of respondents, 65% identified as African-American or Black, 22% as Puerto Rican, and 13% as White or Other. The median age was 42; 68% were male, and 38% were high school graduates. Most individuals were marginally-housed (median stay = 9 months). Almost all participants (96%) paid for medical services via Medicaid. Of the 93% with HIV infection, 44% had been hospitalized at least once in the past year, 72% reported a history of AIDS-defining opportunistic infections, and the median CD4 count was 214. Over two-thirds were actively using drugs and/or alcohol. Among HIV-infected residents, 81% had seen a doctor in the last three months. However, only 67% felt they had a physician, and 48% felt their access to medical care was average to very poor. Among eligible HIV-infected persons, only 39% had taken HAART and 73% had taken PCP prophylaxis in the last week. Predictors for the use of HAART included absence of active cocaine and/or crack use (RR = 3.91; 95% CI 1.03–14.8; p < .03), use of PCP prophylaxis (RR = 5.69; 95% CI .85–38.1; p < .03) and the belief that HAART can help people with AIDS (RR = 1.75; 95% CI 1.28–2.44; p < .03). HAART use did not correlate with site or frequency of medical care or active alcohol or heroin use. Individuals with regular doctors were less likely to have visited an emergency room in the past 3 months (RR = .41; 95% CI .22–.76; p < .02) and more likely to be taking PCP prophylaxis (RR = 2.68; 95% CI 1.19–6.02; p < .008). CONCLUSION: Despite relatively advanced disease in this population of marginally-housed HIV-infected persons, significant proportions do not have a regular primary care provider, are not taking HAART, and report sub-optimal quality of and access to medical care. Active cocaine and/or crack use correlate with a lesser use of HAART.