OBJECTIVE Prior to the widespread use of Mycobacterium bovis, Bacille Calmette-Guerin (BCG), inactivated whole cell mycobacterial vaccines had been shown effective in the prevention of tuberculosis. The present study was conducted to determine the safety and immunogenicity of an inactivated whole cell mycobacterial vaccine in persons with HIV infection.DESIGN Randomized, controlled trial. METHODS A total of 39 HIV-positive patients with prior BCG immunization and CD4 cell counts >/= 200 x 10(6) cells/l were randomized to five doses of inactivated Mycobacterium vaccae (MV) vaccine or control vaccine (CV). Lymphocyte proliferation (LPA) and interferon gamma (IFN-gamma) responses to mycobacterial antigens were assayed at baseline, after three and five doses of vaccine and > 1 year later. Parallel studies were conducted in 10 HIV-negative subjects with prior BCG immunization. RESULTS Among HIV-positive patients, 19 MV recipients had higher LPA and IFN-gamma responses to MV sonicate than 20 CV recipients after three and five doses of vaccine and > 1 year later. LPA responses to Mycobacterium tuberculosis whole cell lysate increased over time in both groups consistent with prior BCG immunization and current antiretroviral therapy; after three doses, responses were boosted to higher levels in MV subjects than CV subjects. LPA responses to WCL were also boosted in HIV-negative MV recipients. Immunization was safe and had no adverse effects on HIV viral load or CD4 cell count. CONCLUSIONS In BCG-primed, HIV-positive and HIV-negative subjects, MV induces durable cellular immune responses to a new mycobacterial antigen and boosts pre-existing responses to WCL. MV is a candidate for clinical trials for the prevention of HIV-associated tuberculosis.