We analyzed the outcome of therapy for 139 patients who were treated for a presumptive diagnosis of pulmonary tuberculosis based on radiographic abnormalities. Patients who had acid-fast bacilli seen on sputum smears and patients who had received adequate therapy for tuberculosis in the past were excluded from the analysis. Accuracy of the diagnosis was determined by comparison of clinical and radiographic findings after 3 months of isoniazid, rifampin, and ethambutol, as well as the results of sputum cultures. Of 139 patients started on therapy presumptively, 66 (48%) were determined to have current tuberculosis (16 had positive cultures, 43 because of improvement in their chest films, and 7 because of clinical improvement). Adverse reactions requiring change of therapy occurred in six of 72 (8.3%) patients determined to have inactive tuberculosis. One patient had both tuberculosis and carcinoma found at bronchoscopy after 3 months of therapy. For purposes of comparison, chest radiographs of 59 patients documented by culture to have current tuberculosis were reviewed. Of these, 45 (70%) were improved at 3 months. Presumptive therapy is of benefit in that it stops progression of the disease at an early stage and decreases the potential for spread of tuberculous infection. In addition, such therapy coupled with systematic reevaluation of patients substantiates the diagnosis or indicates that further evaluation is needed. These benefits must be weighed against the adverse reactions and costs of overtreating patients with inactive disease. Determining the appropriateness of presumptive therapy must be based on local factors including prevalence of tuberculosis and available resources.