Since 1978, we have surveyed the 28 metropolitan health departments initially reporting > 250 cases of tuberculosis/yr to determine the standard of practice in the control of pulmonary tuberculosis and the factors affecting treatment policy. In this survey, results were compared with data obtained in 1978, 1980, 1984, and 1988. As in the previous years, all departments completed the survey. The predominant treatment regimen was 6 months of chemotherapy (64 ± 1.33% of patients) involving isoniazid (I), rifampin (R), and pyrazinamide (Z). Estimated duration of treatment, which had decreased from 20.2 ± 2.1 months in 1980 to 7.58 ± 1.02 months in 1988, increased to 9.34 ± 2.32 months in 1992 (p <0.01). This was attributed to an increase in drug-resistant cases (17 of 25 programs) and to increased incidence of HIV infection during the previous 4 yr. In 1984, HIV infection was estimated to coincide with tuberculosis in 2.54% of all patients, 7.72% in 1988, and 17.42% in 1992. Several other major departures from prior perceived practices were reported. In 1980, 32.1% of all patients were hospitalized initially for tuberculosis treatment, and this number decreased progressively to 17.8% in 1988; in 1992, 34.2 ± 1.32% of patients with tuberculosis were hospitalized for initial treatment. In 1988, no program reported regular use of alternative therapy to isoniazid for chemoprophylaxis; in 1992, 21 programs used alternative regimens (predominantly R-containing). In 1992, nine programs reported increased funds for treatment of tuberculosis (27.2 ± 1.97% after inflation), whereas 16 reported a mean decrease of 14% after inflation. We find that tuberculosis treatment in the major metropolitan health departments in the United States consists predominantly of short-course chemotherapy utilizing I, R, and Z and that overall mortality is not reported to be greater because of initially drug-resistant organisms. However, HIV-associated disease now is a major etiologic factor in tuberculosis, and the number of hospitalizations has doubled in 4 yr. Despite diminished resources, incidence of hospitalization and duration of treatment have increased for the first time in 12 yr. In the absence of an increase in funds for treatment, we conclude that problems currently identified in tuberculosis control by major metropolitan health departments likely will be exacerbated in the immediate future.