Pulmonary disease is sometimes treated empirically as tuberculosis (TB) in the absence of microbial confirmation if the clinical suspicion of active TB is high. In a country of relatively high TB and low HIV burden, we retrospectively studied 107 patients (69.2% male; mean age (SD): 45 (17) years) who received empirical anti-TB treatment for intrapulmonary opacities or pleural effusions suspected of active TB in our hospitals between 1998 and 2002. The diagnosis of definite or probable 'smear-negative' pulmonary TB was made based on treatment outcome at two months with rifampicin, isoniazid, pyrazinamide and ethambutol (or streptomycin). At this end-point, 81 patients (84.4%) had both clinical and radiological improvement (definite cases), 12 (12.5%) had clinical improvement alone and 3 (3.1%) had radiological improvement alone (probable cases). Confirmation of acid-fast bacilli was subsequently obtained in 12 patients (all definite cases) from culture of initial pulmonary specimens. Eleven patients (10.5%) were diagnosed as 'non-TB' based on absence of both clinical and radiological improvement or discovery of another cause for the pulmonary condition at or before this two-month study end-point. In the 'non-TB' group, 2 had carcinoma, 2 had HIV-related pulmonary diseases, 1 had bronchiectasis, while in 6 causes were indeterminate. Six (6.3%) and 3 (27.3%) patients reported adverse effects from anti-TB drugs from the 'TB' and 'non-TB' groups respectively. Our findings suggest that empirical anti-TB treatment is an acceptable practice if clinical suspicion is high in patients coming in our region.