OBJECTIVE To compare the clinical outcomes of early versus late tracheostomy in patients who require prolonged mechanical ventilation. METHODS A prospective observational study was done. The sample was a cohort of 90 patients who had tracheostomy in the medical intensive care unit of a university-affiliated teaching hospital. Primary outcome measures were duration of mechanical ventilation and total cost of hospitalization. Tracheostomy was defined as early if performed by day 10 of mechanical ventilation and late if performed thereafter. RESULTS Fifty-three patients had early tracheostomy (mean +/- SD = day 5.9 +/- 7.2 of ventilation), and 37 patients had late tracheostomy (mean +/- SD = day 16.7 +/- 2.9) (P < .001). The mean (+/- SD) duration of mechanical ventilation was 28.3 +/- 28.2 days in the early-tracheostomy group versus 34.4 +/- 17.8 days in the late-tracheostomy group (P = .005). Total cost of hospitalization was significantly lower in the early-tracheostomy group (mean +/- SD = $86,189 +/- $53,570) than in the late-tracheostomy group (mean +/- SD = $124,649 +/- $54,282) (P = .001). Male sex (adjusted odds ratio = 3.84; 95% CI = 2.32-6.34; P = .007) and higher ratios of PaO2 to fraction of inspired oxygen (adjusted odds ratio = 1.01; 95% CI = 1.00-1.01; P = .03) were associated with early tracheostomy. The timing of tracheostomy was not associated with hospital mortality. CONCLUSIONS Early tracheostomy is associated with shorter lengths of stay and lower hospital costs than is late tracheostomy among patients in the medical intensive care unit. Prospective clinical trials are necessary to determine the optimal timing of tracheostomy in that setting.