Techniques to improve the reliability of expectorated sputum culture were evaluated in 46 patients using a wash technique, quantitative culture, and a combination of washing and quantification. The wash procedure consisted of a jet of tap water over the specimen contained in a tea strainer. The true lower respiratory tract flora was determined by percutaneous transtracheal aspirates, which served as the basis for evaluating the methods of processing expectorated specimens. Either washing alone or quantification alone decreased the number of specimens with organisms that were not present in companion transtracheal aspirates; however, results that were sufficiently improved to be clearly advantageous for clinical interpretation were achieved only when washing and quantification were combined. Washed sputum specimens yielded oropharyngeal "contaminants" in concentrations exceeding 10(6) bacteria per ml in 12 of 46 specimens (26 per cent). Using this technique, only one specimen contained a misleading potential pathogen, and only one specimen failed to yield a potential pathogen that was recovered with the transtracheal aspirate. The wash procedure decreased the mean concentrations of contaminants approximately 100-fold for all specimens and 1,000-fold for purulent specimens.