Complementary and alternative medicine (CAM) use is common in the general population, accounting for substantial expenditures. Among patients with human immunodeficiency virus (HIV) infection, few data are available on the prevalence, costs, and patterns of alternative therapy use. We carried out detailed telephone surveys and medical chart reviews for 289 active patients with HIV in a general medicine practice at a university-based teaching hospital in Boston, Mass. Data were collected on prevalence and patterns of CAM use, out-of-pocket expenditures, associated outcomes, and correlates of CAM use. Of 180 patients who agreed to be interviewed, 122 (67.8%) used herbs, vitamins, or dietary supplements, 81 (45.0%) visited a CAM provider, and 43 (23.9%) reported using marijuana for medicinal purposes in the previous year. Patients who saw CAM providers made a median of 12 visits per year to these providers compared with 7 visits per year to their primary care physician and nurse practitioner. Mean yearly out-of-pocket expenditures for CAM users totaled $938 for all therapies. For the main reason CAM was used, respondents found therapies "extremely" or "quite a bit" helpful in 81 (81.0%) of 100 reports of supplement use, in 76 (65.5%) of 116 reports of CAM provider use, and in 27 (87%) of 31 reports of marijuana use. In multivariable models, college education (odds ratio [OR]=3.7, 95% confidence interval [CI]=1.9-7.1) and fatigue (OR=2.7, 95% CI=1.4-5.2) were associated with CAM provider use; memory loss (OR=2.3, 95% CI=1.1-4.8) and fatigue (OR=0.4, 95% CI=0.2-0.9) were associated with supplement use; and weight loss (OR=2.6, 95% CI=1.2-5.6) was associated with marijuana use. Patients with HIV infection use CAM, including marijuana, at a high rate; make frequent visits to CAM providers; incur substantial expenditures; and report considerable improvement with these treatments. Clinical trials of frequently used CAMs are needed to inform physicians and patients about therapies that may have measurable benefit or measurable risk.