To evaluate the reliability of the OMERACT paediatric ultrasound (US) synovitis definitions and scoring system in JIA.Thirteen sonographers analysed 75 images for the presence/absence of elementary lesions (binary scoring) and for grading synovitis, synovial hypertrophy, effusion and Doppler signals. Static US images of the second metacarpophalangeal joint (MCP-II), wrist, elbow, knee and ankle in JIA patients at different ages and different disease stages were collected with standardized scanning by two experienced sonographers. Intra- and inter-reader reliability were analysed with kappa coefficients.Intra-reader reliability was good for binary scoring (Cohen's kappa 0.62, range 0.47-0.75), synovitis and synovial hypertrophy; excellent for Doppler signals (quadratic weighted kappa 0.77, 0.66-0.86; 0.76, 0.61-0.84; and 0.87, 0.77-0.94, respectively); and moderate for effusion (0.55, 0.24-0.76). Inter-reader reliability was good for synovitis and synovial hypertrophy (Light's kappa 0.68, 95% CI: 0.61, 0.75 and 0.63, 0.54-0.71, respectively), excellent for Doppler signals (0.85, 95% CI: 0.77, 0.90), and moderate for binary scoring and effusion (0.48, 95% CI: 0.36, 0.64 and 0.49, 0.40-0.60, respectively). We obtained the best scores for the knee (0.71, 0.54-0.85) except for Doppler signals, with reliability higher for MCP-II. We found a trend toward better results in older children.This is the first study establishing the reliability of the OMERACT paediatric US synovitis definitions and scoring system in the five most commonly affected joints in JIA. The reliability was good among a large group of sonographers. These results support the applicability of these definitions and scoring system in clinical practice and multicentre studies.