Thrombocytopenia in persons infected with HIV is prevalent and has numerous causes. To study the occurrence, associations, and effect on survival of thrombocytopenia in HIV-infected persons, we used surveillance data from a longitudinal survey of the medical records of 30,214 HIV-infected patients who received medical care from January 1990 through August 1996 in more than 100 medical clinics in 10 U.S. cities. Thrombocytopenia was defined as a physician diagnosis of thrombocytopenia or a platelet count of < 50,000 platelets/ microliter. Analysis of associations of thrombocytopenia was conducted using logistic regression. In HIV+ patients, the 1-year incidence [corrected] of thrombocytopenia was 8.7% in persons with one or more AIDS-defining opportunistic illnesses (clinical AIDS), 3.1% in patients with a CD4 count < 200 cells/mm3 but not clinical AIDS (immunologic AIDS), and 1.7% in persons without clinical or immunologic AIDS. The incidence of thrombocytopenia was associated with clinical AIDS (adjusted odds ratio [AOR] 2.2; 99% confidence interval [CI] 1.7-3.0), immunologic AIDS (AOR 1.5, CI 1.0-2.1), history of injecting drug use (AOR 1.4, CI 1.0-1.9), anemia (AOR 5.0, CI 3.8-6.7), lymphoma (AOR 3.7, CI 1.3-10.6), and black race (AOR 0.7, CI 0.5-0.9). After controlling for anemia, clinical AIDS, CD4 count, neutropenia, antiretroviral therapy, and Pneumocystis carinii pneumonia prophylaxis, thrombocytopenia was significantly associated with decreased survival (risk ratio 1.7; 95% CI, 1.6-1.8). Thrombocytopenia in HIV-infected persons is an important clinical condition associated with shorter survival.