Aim Childhood arthritis arises from several causes. The aim of this observational study is to compare the clinical and biological features and short-term outcome of different types of arthritis because they have different treatment and prognoses. Methods Children <16 years of age hospitalised in a French tertiary care centre for a first episode of arthritis lasting for less than 6 weeks who underwent joint aspiration were retrospectively included. We performed non-parametrical tests to compare groups (septic arthritis (SA), juvenile idiopathic arthritis ( JIA) and arthritis with no definitive diagnosis). The time before apyrexia or C reactive protein (CRP) <10 mg/L was analysed using the Kaplan-Meier method. Results We studied 125 children with a sex ratio (M/F) of 1.1 and a median age of 2.2 years (range 0.3 to 14.6). SA was associated with a lower age at onset (1.5 years, IQR 1.2-3.0 vs 3.6 years, IQR 2.2-5.6), shorter duration of symptoms before diagnosis (2 days, IQR 1-4 vs 7 days, IQR 1-19) and higher synovial white blood cell count (147 cells ×10 3 /mm 3 , IQR 71-227, vs 51 cells ×10 3 /mm 3 , IQR 12-113), than JIA. Apyrexia occurred later in children with JIA (40% after 2 days, 95% CI 17% to 75%) than children with SA (82%, 95% CI 68% to 92%), as did CRP<10 mg/L (18% at 7 days, 95% CI 6.3% to 29.6% vs 82.1%, 95% CI 76.1% to 89.7%, p=0.01). Conclusions There were no sufficiently reliable predictors for differentiating between SA and JIA at onset. The outcomes were different; JIA should be considered in cases of poor disease evolution after antibiotic treatment and joint aspiration.