Seven types of brush catheter were studied in vitro to determine the optimal catheter design for obtaining specimens for bacterial culture using fiberoptic bronchoscopy. The various catheters were inserted through the inner channel, which was contaminated with fresh saliva. The specimen was then obtained by inserting the brush into a broth culture of a marker organism at the distal end of the bronchoscope. Relative merits were based on quantitative bacterial counts of salivary "contaminants" and the marker organism. The catheter that proved superior had telescoping cannulas with a distal occlusion. Twenty-six samplings using catheters with this design uniformly showed no contaminants and high counts of the marker organism. The catheter with telescoping cannulas and a distal occlusion composed of polyethlene glycol was then tested with bronchoscopy performed on 8 healthy volunteers and 6 patients with clinical evidence of a lower airway infection. The specimens yielded likely pulmonary pathogens in high concentrations from patients with pneumonia or lung abscess. Possible contaminants were recovered in low concentrations from 2 of 14 subjects. These in vitro studies and the preliminary clinical trial support the use of this type of catheter for obtaining bronchoscopic specimens for bacterial culture.