STUDY OBJECTIVES: To determine the bronchoscopist's ability to predict specimen quality at the time of transbronchial biopsy and to determine the influence of biopsy specimen size and alveolar content on diagnostic value. DESIGN: Prospective, blinded, observational analysis. SETTING: Tertiary care academic hospital-based pulmonary practice. PATIENTS: Forty-three adult patients who underwent transbronchial lung biopsy. INTERVENTIONS: Each of 170 biopsy specimens was rated as to likelihood of containing diagnostic tissue, size and ability to float, tissue types present, number of alveoli, and pathologic diagnosis. RESULTS: Fifteen percent of biopsy specimens were small and 40% were large. Seventy-six percent of specimens floated; 61.8% of the 170 biopsy specimens contained abnormal lung tissue; and 14.7% of individual specimens were diagnostic. Fifty-two percent of specimens contained >20 alveoli. Larger biopsy specimens were more likely to contain diagnostic tissue (r=0.29, p=0.001). Cup forceps retrieved smaller pieces of tissue (p=0.007) and were less likely to obtain diagnostic tissue (p=0.06). Physician ratings of specimen quality (mean+/-SD) did not differ between specimens containing normal and abnormal tissue (5.98+/-2.3 vs 5.46+/-5.5; p=0.24) or between specimens containing diagnostic vs nondiagnostic tissue (5.56+/-2.5 vs 6.25+/-2.1; p=0.14). Specimens that floated were no more likely to be diagnostic or abnormal than specimens that sank (p