The development of pulmonary infiltrates is an ominous sign in the immunocompromised host (ICH). Selection of the best diagnostic and therapeutic approach is often difficult, and in part depends on the risk-to-benefit ratio of various diagnostic modalities, such as bronchoscopy, bronchioalveolar lavage, percutaneous needle biopsy, and open-lung biopsy (OLB). We reviewed our experience with OLB and bronchoscopy in a predominantly pediatric bone marrow transplantation population, and attempted to assess the frequency with which OLB results directed a therapeutic change, as well as the clinical results of any such therapeutic alteration. A retrospective chart review was conducted of 87 bone marrow transplantation recipients undergoing diagnostic OLB from 1975 to 1986. Bronchoscopic and OLB cultures, histopathologic studies, serological data, and autopsy results were all carefully examined. An assessment of therapeutic alteration as a result of OLB was made, and clinical changes attributable to an OLB-directed therapeutic alteration were sought. Ninety-four OLBs and 37 bronchoscopic examinations were performed in 87 patients. All patients had undergone bone marrow transplantation, most often for leukemia (58/87) or aplastic anemia (13/87). The mean interval from bone marrow transplantation to OLB was 106 days. There were no intraoperative complications, but minor postoperative surgical complications were frequent (incidence, 21%). Postoperative mortality, defined as a death occurring within 30 days of surgery, was 45% (39/87). Seventy-four percent of the patients (64/87) died during the course of the study, at a mean of 43 days after OLB. Most OLBs (60%) yielded a specific diagnosis, defined as the establishment of a precise cause for the infiltrate.(ABSTRACT TRUNCATED AT 250 WORDS)