Background Systemic juvenile idiopathic arthritis (sJIA) is a severe categori of JIA, which includes systemic features such as fever, rash, elevated inflammatory markers along with poly-arthritis. It is frequently resistant to treatment. Until recently, patients were treated mainly with large doses of glucocorticoids. Breakthroughs in understanding the pathogenesis of sJIA have led to the evaluation of biologics, which block the pro-inflammatory cytokines interleukin (IL)-1 or IL-6 and their receptors. Objectives To define the optimal biological agent for sJIA based on safety and efficacy data from randomised controlled trial (RCT). Methods Through a systematic literature search, sJIA RCTs evaluating biological agents were identified. The primary efficacy outcome was defined as 30% improvement according to modified JIA American College of Rheumatology response criteria (JIA ACR30). The primary safety outcome was defined as serious adverse events (SAEs). Outcomes were analysed by pairwise and network meta-analyses. Quality of evidence between biological agents was assessed by applying the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) methodology. Results From the 493 citations original identified, 5 RCTs were eligible for inclusion – one for each of anakinra, canakinumab, and tocilizumab, and two for rilonacept: all vs. placebo. The canakinumab trial distinguished itself from the other trials, as it included patients having high systemic involvement and low joint involvement. Conversely, the tocilizumab trial included patients with more joint involvement compared with most other trials. While all evaluated biological agents were effective, the network meta-analysis indicated with low-quality evidence (due to indirect comparison and inconsistency) that rilonacept-treated patients were less likely to respond than those treated with canakinumab (modified JIA ACR30: odds ratio 0.10 [95%CI 0.02 to 0.38], P =0.001) or tocilizumab (0.12 [0.03 to 0.44], P =0.001). Risks of SAEs were similar among the biological agents (supported by very low-quality evidence) and not different from placebo. Conclusions Despite heterogeneous eligibility criteria and study designs across the 5 studies, and different modified JIA ACR30 criteria, this meta-analysis of short-term RCTs presents empirical evidence that canakinumab and tocilizumab are more effective than rilonacept. Our study also indicated that canakinumab should be considered mainly in patients with high systemic involvement and limited joint involvement, which to some extent also might be true for anakinra, whereas tocilizumab seems to be appropriate in patients with extensive joint involvement. Biological agents in sJIA seem safe and comparable with respect to SAE risk in the short term. We recognize that further data will be needed before these conclusions can be interpreted with high confidence. Protocol registration PROSPERO CRD42013004736 Acknowledgements We wish to thank The Oak Foundation for supporting the Musculoskeletal Statistics Unit, Parker Institute with unrestricted research grants. Disclosure of Interest S. Tarp Grant/research support from: paid to institute: AbbVie, Bristol-Myers Squibb, Mundipharma, and Roche, Speakers bureau: paid to institute: AstraZeneca, Pfizer, and Norpharma, G. Amarilyo Grant/research support from: Novartis, Consultant for: Novartis, I. Foeldvari Consultant for: Abbott, and Chugai, Speakers bureau: Pifzer, R. Christensen Grant/research support from: paid to institute: Abbott, Axellus, Bayer HealthCare Pharmaceuticals, Biogen Idec, Bristol-Myers Squibb, Cambridge Weight Plan, Ipsen, Laboratoires Expanscience, MSD, Mundipharma, Norpharma, Pfizer, Roche, and Wyeth, Consultant for: paid to institute: Abbott, Axellus A/S, Bristol-Myers Squibb, Cambridge Weight Plan, Norpharma, Pfizer, Roche, Mundipharma, and Roche, J. Woo: None declared, N. Cohen: None declared, T. Pope: None declared, D. Furst Grant/research support from: AbbVie, Actelion, Amgen, BMS, Gilead, GSK, NIH, Novartis, Pfizer, Roche/Genentech, UCB, Consultant for: AbbVie, Actelion, Amgen, BMS, Janssen, Gilead, GSK, NIH, Novartis, Pfizer, Roche/Genentech, UCB, Speakers bureau: AbbVie, Actelion, and UCB