From 1988 to 1990, an outbreak of multidrug-resistant tuberculosis (MDR-TB) among patients, and an increased number of tuberculin-skin-test conversions among healthcare workers, occurred on the HIV ward of Jackson Memorial Hospital, Miami, Florida, USA. Measures similar to those subsequently recommended in the 1990 Centers for Disease Control and Prevention guidelines were implemented on the HIV ward by June, 1990, and in September, 1992, we evaluated the efficacy of these control measures. Among MDR-TB patients and healthcare workers with tuberculin-skin-test conversions on the HIV ward, we looked for evidence of exposure to HIV ward MDR-TB patients positive for acid-fast bacilli in sputum during initial (January-May, 1990) and follow-up (June, 1990-June, 1992) periods. Exposure before implementation of control measures to an infectious MDR-TB patient on the HIV ward was recorded in 12 of 15 (80%) MDR-TB patients during the initial period and 5 of 11 (45%) MDR-TB patients during follow-up. After implementation of control measures, no episodes of MDR-TB could be traced to contact with infectious MDR-TB patients on the HIV ward. Skin-test conversions among workers on the HIV ward declined from 7 of 25 (28%) during the initial period to 3 of 17 (18%) in the early (June, 1990-February, 1991) and 0 of 23 in the late (March, 1991-June, 1992) follow-up periods (p < 0.01). Skin-test conversions among healthcare workers were not associated with increased exposure to MDR-TB patients, and were not significantly higher among workers on the HIV ward than on a control ward without tuberculosis patients (3/27 vs 0/16). These data demonstrate that implementation of measures similar to the Centers for Disease Control and Prevention 1990 tuberculosis-control guidelines were effective in halting transmission of MDR-TB to healthcare workers and HIV-infected patients.