OBJECTIVE: To assess the performance of computed tomography (CT) scanning for ascertaining sacroiliitis in patients with suspected spondylarthritis (SpA). METHODS: The Echography in Spondylarthritis French cohort consists of 489 patients with suspected SpA. At baseline, all patients underwent clinical examination, HLA-B typing, and pelvic radiography. Pelvic CT scanning was performed if sacroiliitis on radiography was considered uncertain or if patients presented with buttock pain duration of >6 months. A set of 100 paired radiographs and CT scans was read in a blinded manner by 2 radiologists, and the kappa coefficient was used to assess their interreader reliability. One of the radiologists read the 173 available pairs of radiographs and CT scans performed at baseline. RESULTS: After training, interreader reliability was moderate for sacroiliitis grading on radiographs (κ = 0.59), excellent on CT scans (κ = 0.91), and excellent for ascertaining sacroiliitis on both radiographs (κ = 1) and CT scans (κ = 0.96). The first and second readers considered the quality of imaging to be excellent in 66% and 67%, respectively, of the radiographs (κ = 0.88) and in 93% and 92%, respectively, of the CT scans (κ = 0.93). Concordance between radiographs and CT scans was low for sacroiliitis grading (κ = 0.08) or ascertainment (κ = 0.16). Definite sacroiliitis was ascertained on radiographs in 6 patients (3.5%) (confirmed by CT scans in 4 patients) and on CT scans in 32 patients (18.5%). A history of uveitis was associated with definite sacroiliitis on radiographs (P = 0.04) and CT scans (P