BACKGROUND: Pulmonary disease is the most common reason for presentation and the major cause of death in HIV-infected patients. There has been an evolution in the optimal approach to the investigation of a pulmonary infiltrate in HIV-infected patients since the introduction of induced sputum for the diagnosis of Pneumocystis carinii pneumonia (PCP). AIMS: To evaluate the usefulness of flexible fibreoptic bronchoscopy (FFB), bronchoalveolar lavage (BAL), transbronchial biopsy (TBB) and bronchial brushings (BB) in the diagnosis of pulmonary disease in HIV-infected patients and to examine the effect of FFB on changes in therapy and survival. METHODS: The histories of all HIV-infected patients referred to Fairfield Hospital for FFB between January 1990 and June 1993 were examined retrospectively. RESULTS: Forty-two FFB were performed on 41 patients (40 male and one female). Definitive diagnoses made at FFB included Kaposi’s sarcoma (KS) (n = 9), invasive aspergillosis (n = 5), PCP (n = 4), Mycobacterium avium complex (MAC) pneumonia (n = 2), cytomegalovirus (CMV) pneumonia (n = 1), Cryptococcus neoformans pneumonia (n = 1), microsporidium (n = 1) and Pseudomonas aeruginosa pneumonia (n = 1). TBB and BB did not provide a diagnosis for diseases not seen macroscopically at FFB or diagnosed by BAL. FFB findings altered diagnosis in 21/42 (50%) presentations and changed therapy in 26/42 (62%) cases. CONCLUSIONS: FFB together with BAL altered the working diagnosis and changed therapy in a significant number of patients. TBB and BB should not be routinely performed in all patients as these procedures are of limited value in this setting.