© 2016 The Author(s). Background: In the present study, we compared the incidence of hospitalized infection among children with juvenile idiopathic arthritis (JIA) following initiation of treatment with biologic agents versus methotrexate (MTX). Methods: We used national Medicaid claims data from 2000 through 2010 to create cohorts of children with JIA who were new users of tumor necrosis factor inhibitors (TNFi), anakinra, and MTX (without concurrent biologic agent use) as defined by a 6-month baseline period of nonuse. Because most anakinra users have systemic juvenile idiopathic arthritis (SJIA), we used claims to identify MTX users who likely had SJIA. Among TNFi users, concurrent MTX use was a time-varying covariate. The study outcome was a primary hospital discharge diagnosis of infection. We calculated adjusted hazard ratios (aHRs) to compare infection rates between biologic agents and MTX. Results: We identified 3075 new MTX users (160 with SJIA), 2713 new TNFi users, and 247 new anakinra users. There was no increased risk of infection associated with TNFi monotherapy versus MTX (aHR 1.19, 95 % CI 0.72-1.94) or with TNFi + MTX combination therapy versus MTX (aHR 1.23, 95 % CI 0.69-2.17). Baseline high-dose oral glucocorticoid use (≥10 mg/day of prednisone) was associated with infection (aHR 2.03 [95 % CI 1.21-3.39] versus no oral glucocorticoid). Anakinra was associated with infection versus MTX (aHR 3.53 95 % CI 1.83-6.82), but less so compared with MTX users with SJIA (aHR 2.69, 95 % CI 0.82-8.82). Conclusions: Neither TNFi monotherapy nor TNFi + MTX combination therapy was significantly associated with hospitalized infection compared with MTX. Anakinra was significantly associated with infection, but there was likely residual confounding by disease phenotype.