Guidelines stated that extracorporeal membrane oxygenation (ECMO) may improve outcomes after refractory cardiac arrest (CA) in cases of cardiogenic shock and witnessed arrest, where there is an underlying circulatory disease amenable to immediate corrective intervention. Due to the lack of randomized trials, available data are supported by small series and observational studies, being therefore characterized by heterogeneity and controversial results. In clinical practice, using ECMO involves quite a challenging medical decision in a setting where the patient is extremely vulnerable and completely dependent on the medical team’s judgment. The present review focuses on examining existing evidence concerning inclusion and exclusion criteria, and outcomes (in-hospital and long-term mortality rates and neurological recovery) in studies performed in patients with refractory CA treated with ECMO. Discrepancies can be related to heterogeneity in study population, to differences in local health system organization in respect of the management of patients with CA, as well as to the fact that most investigations are retrospective. In the real world, patient selection occurs individually within each center based on their previous experience and expertise with a specific patient population and disease spectrum. Available evidence strongly suggests that in CA patients, ECMO is a highly costly intervention and optimal utilization requires a dedicated local health-care organization and expertise in the field (both for the technical implementation of the device and for the intensive care management of these patients). A careful selection of patients guarantees optimal utilization of resources and a better outcome.