INTRODUCTION: In development of the American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) rheumatoid arthritis (RA) remission definitions using clinical trials data, one criterion used to compare different remission definitions was whether, compared with those not in remission, those in remission had evidence of later disease stability defined by x-ray and functional status. Validation of the RA remission criteria using observational study data is necessary before recommending their use in practice. METHODS: Using data from those who met RA criteria in the ESPOIR cohort, we matched each person in remission with a person not in remission and then carried out analyses comparing later stability of x-ray and health assessment questionnaire (HAQ) between the two groups. We compared the predictive validity of the same candidate definitions of remission evaluated in the ACR/EULAR process. To minimize potential bias and produce more stable results, we used a bootstrap resampling approach to select those not in remission, repeating the sample matching analysis process 500 times. RESULTS: Results were similar to those of clinical trials analyzed for the ACR/EULAR remission criteria. Specifically, the ACR/EULAR remission definitions using either an simple disease activity index (SDAI) ≤ 3.3, clinical disease activity index (CDAI) ≤ 2.8 or a definition of remission requiring tender joint count, swollen joint count, patient global assessment all ≤ 1 performed as well or better than other candidate definitions of remission in terms of predicting later x-ray and function stability. CONCLUSIONS: ACR/EULAR definitions of remission developed for trials are similarly valid in observational studies in RA and could be used in practice.