Protected samples of lower respiratory tract secretions processed by quantitative culture techniques are recommended to diagnose nosocomial bacterial pneumonia in intubated, mechanically ventilated patients. To evaluate the accuracy of a simple and inexpensive sampling device in this setting, we compared quantitative cultures of paired single-sheathed plugged telescoping catheter (PTC) and protected specimen brush (PSB) samples in 55 patients during 78 suspected episodes of nosocomial pneumonia. PTC and PSB samples were taken in randomized order, and patients were also randomized to have PTC samples taken "blindly" or via a fiberoptic bronchoscope. Fifteen PSB and 27 PTC samples were culture positive (greater than or equal to 10(3) cfu/ml). The two sampling procedures gave similar results in 58 (74%) episodes. A major discrepancy occurred in 20 episodes, including six false negatives of PSB in episodes of proved pneumonia, four possible false positives of PSB, and 10 possible false positives of PTC (three of which rapidly evolved towards overt pneumonia). The sensitivity and specificity of PTC were 100 and 82.2%, and those of PSB were 64.7 and 93.5%, respectively. Blinded or directed PTC samples had similar concordance with PSB samples taken via bronchoscopy. We conclude that PTC is at least as accurate as PSB in the bacteriologic diagnosis of nosocomial pneumonia in intubated patients, and that its use can result in substantial cost savings, especially when fiberoptic bronchoscopy is not otherwise indicated.