The Extracorporeal Life Support Organisation (ELSO) recommends initiating veno-venous extracorporeal membrane oxygenation (ECMO) with sweep gas flow rate (V˙sweep) of 2 L/min and extracorporeal circuit blood flow (Q˙ec) of 2 L/min. We used an in-silico model to examine the effect on gas exchange of initiating ECMO with different V˙sweep and Q˙ec, and the effect of including 5% CO2 in sweep gas. This was done using a set of patient examples, each with different physiological derangements at baseline (before ECMO). When ECMO was initiated following ELSO recommendations in the patient examples with significant hypercapnia at baseline, PaCO2 sometimes fell to < 50% of the baseline PaCO2, a magnitude of fall associated with adverse neurological outcomes. In patient examples with very low baseline arterial oxygen saturation (SaO2), initiation of ECMO did not always increase SaO2 to > 80%. Initiating ECMO with V˙sweep of 1 L/min and Q˙ec of 4 L/min, or with sweep gas containing 5% CO2, V˙sweep of 2 L/min, and Q˙ec of 4 L/min, reduced the fall in PaCO2 and increased the rise in SaO2 compared to the ELSO strategy. While ELSO recommendations may suit most patients, they may not suit patients with severe physiological derangements at baseline.