OBJECTIVE To determine the impact of fiberoptic bronchoscopy (FOB), including quantitative bacterial cultures obtained by BAL and protected specimen brushing on therapeutic decisions and outcome in bone marrow transplant (BMT) patients. METHODS Retrospective review of all BMT patients undergoing FOB during a 4-year period. METHODS A tertiary care university hospital. RESULTS Three hundred five patients underwent BMT; 71 (23%) had FOB to assess pulmonary infiltrates. Allogeneic BMT recipients underwent FOB 3.37 times more often than autologous recipients (p < 0.001). Pathogens were identified in 31 (46%) patients undergoing FOB; bacteria were most commonly isolated although 86% of patients had received broad-spectrum empiric antibiotics. Therapy was changed in 20 (65%) patients when a microorganism was identified and in 9 (22%) with nondiagnostic results (p = 0.0026), but isolation of a presumed pathogen had no apparent effect on survival. There were 19 (27%) FOB complications, including bleeding in 8 (11%) patients and death in 2 (3%). Major complications were associated with prolonged prothrombin time (p = 0.006) and were more common (36% vs 14%; p < 0.05) in patients who had protected specimen brushing vs BAL alone. Mortality at 40 months in BMT patients not requiring FOB was 33% compared with 61% mortality in those undergoing FOB (p < 0.001); mortality was 96% in patients with respiratory failure requiring mechanical ventilation. CONCLUSIONS FOB is diagnostically useful in the evaluation of some BMT patients with pulmonary complications and often influences therapy, although no impact on survival was clearly demonstrated. FOB should be performed only after benefits of the procedure are weighed carefully against its increased risk in this select population.