A conscious sheep model was developed to study sequentially the alveolar milieu by respiratory physiological measurements, transbronchial lung biopsy and bronchoalveolar lavage. The former included all lung volumes, respiratory mechanics and gas exchange measurements. For all these measurements, intraindividual variations were similar to those reported in humans or anesthetized primates. They were not significantly altered by the repetition of each procedure. Transbronchial lung biopsy specimens were adequate in 80% of cases and yielded 100 alveoli and more in 75% of specimens (mean alveolar count = 690; bronchioles = 4.1; vessels = 3.2). Bronchoalveolar lavage (BAL) yielded a 60% effluent and a mean > 9 .times. 106 cells in each of the 3 sets of measurements. BAL induced a mild transient hypoxemia that was completely resolved within 2 h. Cellular and biochemical analyses of BAL effluent in the 3 series done at 2 wk intervals did not change significantly. The procedures repeated on the conscious sheep model at 2 wk intervals do not influence the respiratory physiological measurements, lung histopathology, or BAL analyses.