OBJECTIVES: Patients with moderate-to-severe psoriasis require long-term treatment, yet few clinical trials compare outcomes beyond a short-term induction period. To our knowledge, no network meta-analysis (NMA) of longer-term data has been performed. This NMA aimed to compare longer-term outcomes of currently approved biological therapies and apremilast. METHODS: A systematic review (2000 to August 2016) identified studies reporting Psoriasis Area Severity Index (PASI) 75, 90 and 100 responses. Feasibility of an NMA on maintenance phase PASI endpoints was assessed and sources of heterogeneity considered. Data appropriate for analysis were modelled using a Bayesian multinomial likelihood model with probit link. Wherever possible, study data corresponding to an intention-to-treat approach with non-responder imputation was used. RESULTS: Twenty-two studies reporting outcomes at 40-60 weeks were included, but heterogeneity in study design led to use of a step-wise approach to the synthesis. Four 52-week RCTs were included in the primary indirect comparison, which found brodalumab to be significantly more efficacious than secukinumab (Risk ratio: 1.32 [95% Credible Interval: 1.06, 2.02]), ustekinumab (1.90 [1.26, 3.46]) and etanercept (3.31 [1.58, 10.00]) in terms of PASI 100 response. In secondary analyses, 18 additional studies and four more drugs were included in the network by comparing maintenance phase outcomes from active interventions to induction phase outcomes from placebo arms. For this it was assumed that, had placebo therapy been continued, no change in response would have been observed between end of induction and maintenance. Results were consistent with the main analysis: brodalumab appeared to be most effective, followed by ixekizumab, secukinumab, ustekinumab, infliximab, adalimumab, etanercept and apremilast. CONCLUSIONS: Results of this NMA suggest that brodalumab is associated with the highest likelihood of maintained PASI response after a year of treatment. Further long-term active-comparator RCT data is required to better assess relative efficacy across the range of currently approved therapies.