Background: Venovenous extracorporeal membrane oxygenation (ECMO) has become the ultimate supporting technique for the rescue of severe acute respiratory distress syndrome (ARDS) patients. Although tracheostomy during ECMO has proven to be beneficial, the proper time point for performing the tracheostomy remains unclear. The purpose of our study was to demonstrate whether early tracheostomy (ET; within 7 days of ECMO initiation) outweighs delayed tracheostomy (DT; 8 days of more after ECMO initiation).Methods: A retrospective cohort study was established. All ARDS patients who underwent tracheostomy during V-V ECMO support in the intensive care unit (ICU) of a tertiary hospital from December 2013 to November 2020 were reviewed.Results: Of the 187 ARDS patients who received V-V ECMO support, 30 (16%) underwent tracheostomy—18 (60%) during ECMO support and the other 12 after ECMO decannulation. Among the 18 patients who underwent tracheostomy while receiving ECMO, 11 (61.1%) received ET, and 7 received DT. No significant difference was found between the ET and DT groups in terms of demographic data, medical history, disease severity (estimated based on the RESP, PRESERVE, APACHE Ⅱ, SOFA and Murray scores), ARDS risk factors or mechanical ventilation duration before ECMO. The ET group showed a decreased incidence of ventilator-associated pneumonia (VAP) during ECMO support (45.5% vs. 100%; P= 0.038) and shortened durations of ECMO (9.0 vs. 27.0 days; P= 0.011) and mechanical ventilation (16.0 vs. 56.0 days; P= 0.027). ET did not significantly alter the all-cause ICU mortality rate (54.5% vs. 28.6%; P= 0.367), all-cause hospital mortality rate (which was the same as the ICU mortality rate), length of ICU stay (336 vs. 627 hours; P= 0.085), or length of hospital stay (26 vs. 37 days; P= 0.285). Local bleeding at the tracheostomy wound did not differ between the two groups (27.3% vs. 42.9%, P= 0.627) .Conclusion: Compared with delayed tracheostomy, ET performed within 7 days of ECMO cannulation for severe ARDS patients could decrease the VAP incidence during ECMO support and shorten the durations of ECMO and mechanical ventilation; However, it may not improve the outcome. Prospective and multicenter studies are needed for further research.