OBJECTIVE: To analyze the data on patients operated on for pulmonary tuberculosis (TB) with (Group I) or without (Group II) a correct TB diagnosis and preoperative anti-TB treatment. METHODS: Between 1980 and 1997, 144 resections for TB (Groups I+II) were performed. The 80 patients in Group I underwent therapeutic resections: 32 cases involved recurrent cavities or tuberculomas, three involved post-TB bronchiectasis, 13 involved progression of cavities or tuberculomas, and 32 involved persistent tuberculomas after 6 months of anti-TB therapy. The 64 patients in Group II were operated on for a suspicion of malignancy in 49 cases, for cavitary lesions with haemophthysis in six cases, for multiple lesions in seven cases, and for recurrent hydrothorax in two cases. RESULTS: Groups I and II included 0 and five pneumonectomies, 32 and 29 lobectomies, 48 and 20 wedge resections, 0 and nine videothoracoscopic biopsies, and 0 and one hilar lymphadenectomy, respectively. In Groups I and II, the mean duration of postoperative hospitalization was 13.2 and 10.4 days, and the frequency of postoperative pneumothorax was 11.25 and 4.6%, respectively. The incidence of bronchopleural fistula was 1.25 and 0%, the mortality was 0 and 3.1%, and the morbidity was 53.7 and 35.9% in Groups I and II, respectively. Two patients with active disease died in Group II. Pathology demonstrated that the frequency of acid-fast bacilli in Groups I and II was 40 and 25%, respectively. CONCLUSIONS: Patients without a correct preoperative TB diagnosis underwent more extensive parenchyma resection. Postoperative complications increased when acid-fast bacilli were present. The lack of preoperative anti-TB treatment did not involve a higher risk of minor complications, but death occurred only in this group.