BACKGROUND: To describe the rate and timing of DMARD start in patients with early inflammatory arthritis in France, and to determine the factors leading to this treatment start. METHODS: The ESPOIR cohort study collects data on patients presenting with early arthritis. Baseline characteristics were assessed, and Cox regression analysis was performed to estimate the likelihood of starting DMARD treatment over time, adjusting for patient-, disease- and physician characteristics. RESULTS: Of the 775 analysed patients, 598 (77.2%) received at least 1 DMARD during the follow-up period, after a median time of 4.0 months. In general, a higher tender joint count, involvement of the hands, involvement of more than 3 joint groups, presence of abnormal CRP-levels or CCP-antibodies significantly increased the likelihood of being treated (p<0.01 for all determinants), as well as a positive result on the bilateral foot-squeeze test (p<0.04). In addition, a significant hetero-geneity in therapeutic strategy across the 14 tested French regions was found: adjusted hazard ratios for DMARD start ranged from 1 to 2.15 (p<0.01), depending on the region where a patient was followed. For anti-CCP test and swollen joint count we demonstrated a statistically significant interaction with geographic region, implying that these tests are interpreted differently across regions. The same factors that increased the likelihood to start a DMARD were related to an earlier start. CONCLUSION: Rate and timing of treatment start with DMARDs in patients with early inflammatory arthritis in France is determined by well known clinical and biochemical variables. Apart from these variables, however, unknown and intangible factors that seem to cluster geographically are responsible for important variations in practice performance.