OBJECTIVES: To assess the prevalence of remission in early arthritis, to evaluate the concordance across different criteria sets in defining this state, and to look for predictive factors for early and sustained remission. METHODS: Patients from the ESPOIR cohort were followed-up every 6months. We analysed early remission and sustained remission in 3 groups of patients: patients having rheumatoid arthritis (RA) according to 2010 ACR/EULAR criteria, undifferentiated arthritis (UA), and the whole cohort. Remission was defined according to ACR/EULAR criteria, 28 Joint Disease Activity Score (DAS28<2.6), and Simplified Disease Activity Index (SDAI≤3.3). Agreement was evaluated by k-coefficient. Predictive factors for sustained remission at 1, 3 and 5year in RA patients were analyzed. RESULTS: Eight hundred and nineteen patients were included. Early remission rates in the RA/UA/ESPOIR groups were observed in respectively 29.2% (181/682), 51.4% (55/123) and 32.7% (239/813) of patients by DAS28; 15.7%, 29.1% and 18% by SDAI; and 11.2%, 29.1% and 12.8% by ACR/EULAR criteria. Agreement between classifications of remission was low for DAS28 vs. ACR/EULAR (k=0.44), high for SDAI vs. ACR/EULAR (k=0.78), and moderate for SDAI vs. DAS28 (k=0.54). Lower baseline disease activity scores, non-menopausal status and younger age were the best predictive factors for sustained remission, with consistent results across the 3 definitions of remission. CONCLUSION: Our study showed that the rate of early and sustained remission in early arthritis is dependent on the definition used, with a variable degree of agreement across criteria sets, but with consistent predictive factors of favourable outcome in patients finally diagnosed with RA.