Ventilator-associated pneumonia is frequently diagnosed with quantitative cultures of samples obtained by bronchoscopic techniques, a method associated with high costs and potential adverse effects. Quantitative cultures of endotracheal aspirates are easier and cheaper to obtain, and good correlations between the results of this method and those of bronchoscopic methods have been reported. However, the reproducibility of quantitative cultures of endotracheal aspirates has never been determined. We studied the quantitative analysis of endotracheal aspirates from 21 mechanically ventilated patients taken during two study days with 2- and 6-h intervals between samplings. In all, 40 endotracheal aspirates were obtained. For mechanically ventilated patients, the median variation of quantitative culture results was 12.3% (range, 0 to 63%), corresponding to 0.7 log CFU/ml. Furthermore, variation was independent of the interval of time between samplings. Persistence of significant numbers of pathogens in quantitative cultures (> or = 10(5) CFU/ml) of the consecutive endotracheal aspirates occurred in 82% of samples. We conclude that results of quantitative cultures from endotracheal aspirates are reproducible and may be useful in diagnosing ventilator-associated pneumonia.