The aim of this study was first, to evaluate the value of cough following tracheal tube cuff deflation, and second, to reassess the value of the cuff-leak test to predict postextubation stridor (PES). In spontaneously breathing patients, immediately before extubation, the tracheal tube was deflated and the absence of cough was monitored. The tube was then obstructed with a finger, and the absence of leak was monitored. Extubation was then performed. Four PESs were observed after 115 extubations (incidence: 3.5%). The absence of cough was more frequently observed when PES occurred than when it did not (75% v 21%, P =.04). The absence of leak was observed in 100% of PES and in 20% of PES free extubations (P =.01). The absence of both leak and cough was more frequently observed in PES (75% v 7%, P <.0001). In the absence of leak, the likelihood ratio of developing PES was 5.04 and rose to 10.4 when cough was also absent. The likelihood ratio of not developing PES in the absence of leak alone was 0. We conclude that in a population of medical intensive care unit spontaneously breathing patients, just before extubation, the presence of leaking around the endotracheal tube rules out PES, whereas the absence of cough and of leak are good predictors of PES.