Background: Although previous work that considered a variety of chronic conditions has shown that higher quality physician-patient communication care is related to better health outcomes, the quality of physician-patient communication itself for patients with HIV disease has not been well studied. Objective: To determine the relationship of patient, visit, physician, and physician practice characteristics to two measures of physician-patient communication for patients with HIV disease. Design: Cross-sectional survey of physicians and patients. Setting: Cohort study enrolling patients from throughout eastern Massachusetts. Study subjects: 264 patients with HIV disease and their their primary HIV physicians (n = 69). Measurements: Two measures of physician-patient communication were used, a five-item general communication measure (Cronbach's α = 0.93), and a four-item HIV-specific communication measure that included items about alcohol, drug use, and sexual behaviors (Cronbach's α = 0.92). Results: The mean age of patients was 39.5 years, 24% patients were women, 31.1% were nonwhite, and 52% indicated same-sex contact as their principal HIV risk factor. The mean age of physicians was 39.1 years, 33.3% were female, 39.7% were specialists, and 25.0% self-identified as gay, lesbian, or bisexual. In multivariable models relating patient and visit characteristics to general communication, longer reported visit length (p < .0001), longer duration of the physician-patient relationship (p = .02), and female gender (p = .04) were significantly associated with better communication. The interaction of patient gender and visit length was also significant (p = .02); longer visit length was more strongly associated with better general communication for male than female patients. In similar models relating patient and visit characteristics to HIV-specific communication, longer visit length (p < .0001) and less advanced disease stage (p = .009) were associated with better communication. In multivariable models relating physician and practice characteristics to general communication no variables were significant. However, both female physician gender (p = .002) and gay/lesbian/bisexual sexual preference (p = .003) were significantly associated with better HIV-specific communication. Conclusions: In this study, female and homosexual physicians provided higher quality HIV-specific communication than male and heterosexual physicians. Better understanding the processes by which female and homosexual physicians achieve higher quality communication may help other physicians communicate more effectively. Health care providers and third-party payers should be aware that shorter visits may compromise physician-patient communication, and that this effect may be more consequential for male patients.