Background Following the US transition from ICD-9-CM to ICD-10-CM in 2015, the number of codes for chronic inflammatory arthritis increased from 9 to 193. It is unclear how this transition affected diagnoses of juvenile idiopathic arthritis (JIA) within claims databases. Objectives Determine how the ICD transition affected the identification of incident JIA diagnosis, including by JIA type and treatment. Methods In this retrospective cohort study, eligible members were commercial health plan members from the HealthCore Integrated Research Database <18 years old enrolled January 2006 to February 2020 with ≥6 months of enrollment and no prior rheumatic disease or immunosuppression. Diagnosis of JIA was based on ICD-9-CM algorithms previously used or validated in other claims data. Initial cohort identification was performed using a Sentinel Initiative modular program. Poisson and negative binomial regressions adjusting for calendar time, month post-switchover, ICD era, and seasonality were performed. Incidence rates (95% confidence intervals) are expressed per 100,000 members per month. Results We identified 6260 and 2308 incident JIA diagnoses using ICD-9-CM and ICD-10-CM codes, respectively. Mean age at diagnosis was 12 years (SD 4.8); differences by sex (64% female) and JIA type were consistent with known JIA epidemiology. The monthly incidence of diagnosed JIA did not change within either coding era but declined abruptly at the ICD-9/10-CM transition from 1.57 (1.55, 1.59) to 1.33 (1.30, 1.37). When stratified by JIA type, incidence declined for seronegative poly- or oligoarticular JIA [1.21 (1.19, 1.22) to 0.93 (0.90, 0.95)] and juvenile spondyloarthritis [0.27 (0.27, 0.28) to 0.23 (0.22, 0.24)] but doubled for systemic JIA (0.06 to 0.12); the incidence of seropositive JIA was slightly increased [0.052 (0.051, 0.053) to 0.067 (0.065, 0.069)]. The incidence of JIA without attendant drug claims declined (33% to 27%), whereas incident JIA treated with biologics within 6 months of diagnosis increased abruptly (25% to 35%), despite no market changes around the transition. Among members diagnosed with JIA by ICD-9-CM but not ICD-10-CM (N=1270) despite continued enrollment after 2015, common alternate diagnoses included anxiety, fatigue, non-specific musculoskeletal pain, and headaches. Conclusions The incidence of diagnosed JIA, both overall and without expected medical treatment, declined abruptly at the US ICD-9/10-CM transition, with apparent reclassification of some persons to diagnoses consistent with chronic pain disorders. Increases in certain JIA subgroups were noted, including systemic JIA and biologic-treated JIA, suggesting improved diagnostic specificity with ICD-10-CM codes.