Objective. The most widely used classification criteria for SLE are those derived and validated in adult patients by the ACR. Alternatives include the Boston weighted (BW) and SLICC criteria. The aim of this study was to compare the performance of BW and SLICC criteria with the 1997 ACR criteria in a JSLE cohort. Methods. Cases were JSLE patients and controls were patients with other rheumatic diseases attending a tertiary centre in the past 10 years. Data were retrospectively collected to establish the ACR, BW and SLICC criteria fulfilled at the first visit and within the first year of follow-up. A consensus diagnosis of JSLE established by the same group of highly experienced paediatric rheumatologists was chosen as the standard of reference. Results. One hundred and seventy-three patients were included: 81 JSLE and 92 controls. There was a sharp increase in sensitivity and prevalence of all criteria within the first year of follow-up. The BW criteria had higher sensitivity than the ACR criteria (81.5% vs 58%, P<0.001) at the first visit, but lower specificity in both periods. SLICC criteria had higher sensitivity (82.7% vs 58%, P<0.001) at the first visit, but similar specificity in both periods. Conclusion. In this JSLE population, the SLICC criteria performed best in terms of sensitivity and accuracy at the first visit and within the first year of follow-up.