Background Since March 2020, health care systems were importantly affected by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak, with some patients presenting severe acute respiratory distress syndrome (ARDS), requiring extra-corporeal membrane oxygenation (ECMO). We designed an ambispective observational cohort study including all consecutive adult patients admitted to 5 different ICUs from a university hospital. The main objective was to identify the risk factors of severe COVID-19 ARDS patients supported by ECMO associated with 90-day survival. Results Between March 1st and November 30th 2020, 76 patients with severe COVID-19 ARDS were supported by ECMO. Median (interquartile range IQR) duration of mechanical ventilation (MV) prior to ECMO was of 6 (3–10) days. At ECMO initiation, patients had a median PaO2:FiO2 of 71 mmHg (IQR 62–81), median PaCO2 of 58 mmHg (IQR 51–66) and a median arterial pH of 7.33 (IQR 7.25–7.38). Forty-five patients (59%) were weaned from ECMO. Twenty-eight day, 60-day and 90-day survival rates were, respectively, 92, 62 and 51%. In multivariate logistic regression analysis, with 2 models, one with the RESP score and one with the PRESERVE score, we found that higher BMI was associated with higher 90-day survival [odds ratio (OR): 0.775 (0.644–0.934), p = 0.007) and 0.631 (0.462–0.862), respectively]. Younger age was also associated with 90-day survival in both models [OR: 1.1354 (1.004–1.285), p = 0.044 and 1.187 (1.035–1.362), p = 0.014 respectively]. Obese patients were ventilated with higher PEEP than non-obese patients and presented slightly higher respiratory system compliance. Conclusion In this ambispective observational cohort of COVID-19 severe ARDS supported by ECMO, obesity was an independent factor associated with improved survival at 90-day.