Objectives To assess the impact of tracheostomy timing on outcome of critically ill patients requiring mechanical ventilation (MV). Study design Retrospective clinical study in a twelve beds intensive care unit (ICU). Patients and methods From January 2001 to June 2005, patients under MV who received tracheostomy were divided into 2 groups: early tracheostomy group when tracheostomy was performed before or on day 7 and late tracheostomy group when it was performed thereafter. We compared prevalence of nosocomial pneumonia, length of sedation, lengths of MV, length of stay in ICU, weaning from MV and mortality rates between the 2 groups. Results During this period of 4 years and half, 112 patients underwent tracheostomy, 62 of whom had early tracheostomy and 50 had late tracheostomy. Early tracheostomy was associated with significant reduction of length of sedation (10 ± 3 vs 17 ± 5 days, P < 0.001), length of MV (21 ± 19 vs 29 ± 17 days, P = 0.02) and length of stay in ICU (33 ± 22 vs 42 ± 18 days, P = 0.042). There were no differences in prevalence of pneumonia (21% for early tracheostomy group vs 31% for late tracheostomy group, P = 0, 13), weaning from MV (50 vs 36%, P = 0.19), and mortality rates between the 2 groups (38 vs 54%, P = 0.15). Conclusion This study demonstrated that early tracheostomy (≤ 7 days), was associated with shorter length of sedation, shorter duration of MV and shorter ICU length of stay, without affecting weaning from MV, prevalence of nosocomial pneumonia or survival.