SummaryThe authors analyse the bacteriological data gathered by 100 successive tracheal Punctures and compare these results to those obtained by sputum cultures, which had either been collected by routine or when withdrawing the catheter for tracheal aspiration. This plain and not hazardous technique allows to draw the following conclusions :1) The culture of routinely collected sputum at the patient’s bed-side often misleads the physician (6 times on 10) whereas newly expectorated sputum brought immediately to the laboratory more truly shows the tracheo-bronchic flora (valuable results in 75 % of the cases).2) The tracheal puncture, which reduces the causes of errors, due to the contamination of the sputum by the rhino-pharyngeal flora, to a minimum, particularly is indicated in the following cases : patients who do not expectorate (including suspects of pulmonary tuberculosis), instantly earnest pneumopathia, bad response to applied antibiotic therapy, tests of the true efficiency of an antibiotic therapy.3) The bacteriological study of these punctures performed on patients, who caused therapeutical problems and who had been submitted to antibiotic therapies evidenced the following data : 40 % of the punctures were sterile, 39 % showed one single germ (18 Gram negative, 16 Gram positive, 2 BK, 3 aspergillus), 15 % showed 2 simultaneous germs and 6 % were not significant.