The use of extracorporeal membrane oxygenation (ECMO) as a resuscitative measure during or after manual cardiopulmonary resuscitation (CPR) shows sharply contrasting results. To assess the added value of ECMO in this situation and looking for predictors of mortality we performed a meta-analysis of individual patients collected from observational studies. An electronic Pubmed search restricted to English language publications between 1990 and 2007 using a consensus restrictive criterion retrieved 462 titles. Of those, 93 abstracts were considered appropriate for full text evaluation with 37 articles being included in our meta-analysis. In addition, unpublished data on a series of 98 non-duplicated patients from the author of one of the included studies was added. Data on 288 individually identified patients with a median age of 0.50 years and a median weight was 4.5 kg and demonstrated an overall survival to hospital discharge of 39.6% (114/288). Neurological complications were common, affecting 27% of all patients (77/288) and 14% of those discharged alive (16/114). Other common complications were renal failure (25%) and sepsis (17%). Odds ratios for mortality were higher for the presence of: any complication (OR 3.9, 95% CL 2.3-6.4), neurological (OR 3.3, 95% CL 1.7-6.1), renal (OR 5.1, 95% CL 2.5-10.3) and when the implementation of ECMO took >30 min (OR 2.1, 95% CL 1.1-3.8). Neck vessels cannulation had a lower association with mortality (p<.001). Simple rate comparison between manual CPR alone and the use of emergency ECMO shows a difference on survival to discharge of 12-23%. Its effectiveness is higher when implemented in the first 30 min after arrest. Age and weight do not seem to influence mortality. The incidence of complications is high, particularly neurological and renal, having a strong influence on survival. The specific characteristics of the neurological complications and their long-term effects on survivors are poorly reported in the literature.