Since 1980, we have surveyed at 4-yr intervals the metropolitan health departments initially reporting > 250 cases of tuberculosis to determine the perceived standard of practice for tuberculosis control and the factors affecting formulation of treatment policies. Between 1992 and 1996, use of supervised short-course (6 to 9 mo) intermittent therapy with multiple drugs including isoniazid, ethambutol, pyrazinamide, and rifampin increased from 4.3% to 46% of all new patients. Pyrazinamide use for initial treatment for children has increased substantially and now predominates (74.2% of patients in 1996 versus 48.1% of patients in 1992). Duration of treatment, which was 20 +/- 2.1 mo in 1980, is now 8.00 +/- 2.29 mo in 1996. The incidence of human immunodeficiency virus-associated tuberculosis, which was virtually unrecognized in 1984, has remained the same between 1992 and 1996 (18.0%). As in previous years, there was a wide variance among health departments in the incidence (< 5% to > 40%) of HIV-associated tuberculosis. After years of funding decreases, there has been an impressive increase in resources in the past 4 yr. In 1988, mean budget allocation for health departments decreased by 7.9% versus the prior 4 yr and, in 1992, there was no overall change in budget allocation after inflation versus 1988. In 1996, however, funds for treatment increased by 84 +/- 33%. This increase in funding has been translated into the greatly expanded use of supervised intermittent therapy and aggressive screening programs, which likely have resulted in the decreased incidence of tuberculosis since the prior survey.