OBJECTIVE: To assess ultrasonography's (US) performance to detect the structural damage in the initial evaluation of early arthritis (EA) using the Etude et Suivides Polyarthrites Indifférenciées Récentes (ESPOIR) cohort. METHODS: ESPOIR is a French, multi-centric EA cohort. Four centres assessed the structural damage by both X-ray and US examination at baseline. X-rays of hands and feet were read first by the centre's local investigator (usual reading), then in the X-ray coordinating centre (central reading). Four trained examiners performed US blindly from clinical data to detect erosions on the second and fifth MCP (MCP2 and 5) and the fifth MTP (MTP5) joints bilaterally. RESULTS: Patients' characteristics (n = 126) were: female 78%; mean age 50.3 years; disease duration 103 days; disease activity score on 28 joints 5; CRP level 22.7 mg/l; and 79.4% of the patients fulfilling RA ACR criteria. Twelve patients had missing data for X-rays. US revealed 42 (36.8%) patients with erosive disease, whereas radiography revealed only 30 (26%) with central reading and only 11% with usual reading. US missed erosive disease present in X-rays in 10 (8.8%) patients. Combined technique of both revealed 52 (45.6%) patients with erosive diseases. On the targeted joints, US detected erosion on 75 (11%) joints vs X-rays on only 11 (1.5%). Only three joints with erosion(s) detected on X-rays were missed on US. At baseline, the presence of PD activity was not associated with joint erosions. CONCLUSIONS: US on six joints detected 1.4-fold more patients with erosions (3.3-fold more with the usual reading). In clinical practice, US combined with X-rays is of helpful diagnostic value in EA.