Objective: Comparison of different diagnostic criteria for early liver allograft dysfunction (EAD) and their capability to predict mortality., Design: Single-center, prospective, cohort study., Settings: ICU in a Regional Hospital with a liver transplant program since 1997., Patients: 253 consecutive patients admitted to our ICU immediately after liver transplantation between 2009 and 2015., Variables of Interest: Differences in the incidence of EAD and its relation with ICU, Hospital and 2-year mortality depending on the definition applied using as comparator the UNOS (United Network for Organ Sharing) primary non-function criterion., Results: The incidence of early liver allograft dysfunction according to UNOS was 13.8%, to Makowka 6.3%, to Ardite 10.7%, to Nanashima 20.6%, to Dhillon 30.8% and to MEAF 13.4%. Kappa test did not show a good correlation among these criteria. EAD was related with ICU mortality for all diagnostic criteria except Dhillon but only UNOS, Makowka and MEAF were associated with 2-year mortality. Hospital mortality was poorly predicted by all criteria except for the MEAF score., Conclusion: We found a poor agreement between different criteria analyzed for the diagnosis of EAD. In our population, the MEAF score showed the best relationship with short- and long-term mortality., (Copyright © 2018 Elsevier España, S.L.U. y SEMICYUC. All rights reserved.)