The referral centre of tuberculosis in the municipality of Copenhagen, Denmark was the setting for this study, which aimed to assess the diagnostic strategy (chest X-ray and clinical mycobacteriology) in pulmonary tuberculosis. Patient records and chest X-rays were examined for all patients who had sputum or gastric lavage examined for Mycobacterium tuberculosis (Mtb) from 1 January 1992 to 30 April 1994. All chest X-rays were re-evaluated by a trained lung specialist, who did not know the results of sputum culture. Evaluation was referred to one of seven X-ray categories, and compared to the results of culture. Culture of sputum or gastric lavage were positive for Mtb in 54 (14%) of 392 patients; in 61% of 59 patients with X-ray changes thought to be due to tuberculosis (TB); in 20% of 51 patients with X-ray changes compatible with TB; in 14% of 35 patients with previous TB and radiographically active TB; in 2% of 103 patients with previous TB, but not radiographically active TB; in 1% of 112 patients with X-ray changes thought to be due to other disease; and none out of 32 patients with normal X-ray. Even in this highly selected material, it is relatively expensive to find the very few cases of active TB in patients with chest X-ray changes not suspected to be due to TB. It is recommended that: (1) examination of sputum for Mtb should always be preceded by X-ray of the chest in a low-prevalence country; (2) routine culture of sputum for Mtb is restricted to patients with X-ray changes typical or compatible with active TB; and (3) exceptions to this general rule should be made on the basis of the individual's clinical history.