The aim of this retrospective methodology study was to investigate the influence of using different definitions for restoration failure and inclusion criteria on restoration longevity expressed in AFR.EPF from fifteen general dental practices were used for collecting the data for this study. From the EPF, 321,749 composite restorations placed in 52,245 patients by forty-seven GDPs between January 2000 and December 2011 were included. Kaplan-Meier statistics were applied and mean AFRs over 2, 5 and 10 years were calculated. The effect on the AFR of using different levels of failure: based on Claims data (CD), Success (SUC), Survival (SUR) and different inclusion criteria of tooth/restoration variables were reported.Highest AFRs were found for level CD, in which every intervention was considered as failure, and the lowest AFRs for level SUR in which repairs and an endodontic treatments were not considered as a failure. AFRs increased when the observation period prolonged especially for SUR, followed by SUC and CD. An overview of long-term survival studies showed a wide variation in study design, performed clinical examination (USPHS criteria or GDP), number of restorations included, description of restoration failure and found AFRs for CD, SUC and SUR.Using failure criteria, Success and Survival, in future clinical studies would enable a better comparison of studies as well as demonstrate the impact of more conservative restorative intervention protocols on patient care.