OBJECTIVE: To compare rheumatologists' prescription for first disease modifying antirheumatic drug (DMARD) in early rheumatoid arthritis (RA) in real-life settings with 2 clinical practice guidelines (CPG), the French Society of Rheumatology/STPR 2004 and EULAR/ESCISIT 2007, and thus assess the gap between practices and guidelines. Method. ESPOIR was a French multicenter cohort study of 813 patients with early arthritis between 2002 and 2005. "Definite" and "probable" RA were defined according to ACR criteria and the level of diagnostic certainty. The objectives were to assess conformity between the observed first-line DMARD prescribed for those patients and the DMARD recommended in the guidelines; and to conduct a mail survey of patients' usual rheumatologists to investigate the reasons for their nonconformity with guidelines. RESULTS: In total 627 patients with definite or probable RA were identified. Conformity rates were 58% for STPR guidelines and 54% for EULAR guidelines. At 6 months, 83 (34%) patients with early RA did not receive any DMARD. Main determinants associated with conformity to guidelines were disease activity and presence of severity-predictive factors. The main reason leading to a discrepancy between guidelines and daily practice appeared to be diagnostic uncertainty, i.e., the difficulty to reliably assess RA diagnosis as early as the first visits to the rheumatologist. CONCLUSION: There is a substantial gap between CPG and rheumatologists' daily practice concerning the first DMARD to prescribe in early RA. This is explained mainly by diagnostic uncertainty. More attention should be paid in future guidelines to the diagnostic difficulties of early RA.