Background Management of children with necrotizing enterocolitis (NEC) remains challenging. Various scores try to facilitate therapeutic decision-making. We aim to assess the agreement of three scores intending to predict the need for surgery and/or mortality in our patient cohort, and analyze agreement between the different scores. Methods This study is a retrospective analysis of patients with NEC Bell's stage II and III, managed in a single institution (1991-2011). Three existing scores (Metabolic Derangement Acuity score, NEC score, Detroit score) were calculated individually for each patient. The agreement between predicted outcome by scores and real outcome was evaluated with kappa statistic. Results Of 57 children, 46% presented with NEC stage II, 54% with stage III, 46% were treated with surgery, 54% conservatively, and survival was 58%. The kappa indexes for "need for surgery" were 0.41, 0.13, and 0.12 and kappa indexes for "mortality" were 0.27, 0.04, and 0.1 for the Metabolic Derangement Acuity score, the NEC score, and the Detroit score, respectively. Conclusion In our cohort, the agreement between the predicted outcomes by scores and the real need for surgery and/or mortality was poor. There was a lack of clinical usefulness of the tested scores. We must continue to better identify parameters to help guide the management of these patients.