Brief measures of health-related quality of life are being used with increased frequency in AIDS clinical trials. Self-administration of questionnaires can reduce costs in this setting because they require little time. However, the equivalence between self- and interview-administered responses in clinical trials is not known. We evaluated patient and proxy responses to the Medical Outcomes Study HIV Health Survey (MOS-HIV) and the EuroQol. We randomized 68 patients with advanced HIV disease on (1) mode of administration (self vs. interview); (2) type of interview (face-to-face vs. telephone); (3) questionnaire order (MOS-first vs. EuroQol-first); and (4) 2- vs. 3-item response categories for physical limitations. There were few differences in scores between self and interview administration and type of interview. Proxy respondents viewed patients as more impaired than did patients themselves on subjective aspects of health including mental health (63.8 vs. 75.7, p < 0.001), health distress (67.3 vs. 77.1, p = 0.007), pain (64.4 vs. 70.0, p = 0.04), and vitality (48.4 vs. 55.5, p = 0.04). Results concerning questionnaire order and number of response categories were not conclusive. Our results suggest that for patients with advanced HIV disease, data from the MOS-HIV and the EuroQol collected using different modes may be pooled, but that proxy responses should be calibrated.