OBJECTIVES: To assess the relative efficacy of ibrutinib (IBR), a first-in-class BTK- inhibitor, versus Idelalisib+ofatumumab (IDEL+OFA) and physician’s choice in relapsed/refractory (R/R) CLL-patients using Bayesian Network Meta-Analysis (NMA). METHODS: Three RCTs in R/R CLL-patients were identified with OFA as common treatment arm. IBR (Byrd, 2014) and IDEL+OFA (Jones, 2015) showed improved PFS (HR=0.13 (INV) and 0.27 (IRC), respectively) and OS (HR=0.39 and 0.74, respectively) versus ofatumumab in R/R CLL-patients.  Osterberg (2014) compared PFS (INV) (HR=0.56) and OS (HR=0.72) for OFA to physician’s choice (PC), a mix of well-established CLL-treatments, in more severe patients.  A Bayesian NMA was conducted in line with NICE guidelines, using a fixed-effect model with non-informative priors. Posterior distributions for the HR were summarized by median values and 95% credible intervals. RESULTS: HR for PFS comparing IBR vs IDEL+OFA and PC were 0.49 ([0.28;0.87],P(HR<1)=99.2%) and 0.07 ([0.04;0.13],P(HR<1)=100%), respectively. HR for OS comparing IBR vs IDEL+OFA and PC were 0.52 ([0.24;1.14],P(HR<1)=94.7%) and 0.28 ([0.13;0.80],P(HR<1)=100%), respectively. DISCUSSION : In absence of head-to-head trials, indirect comparisons can provide useful insights to clinicians and reimbursement-decision making on relative efficacy of treatments. The probabilistic interpretation of Bayesian results suits these purposes, allowing probabilistic statements on which treatment is likely to be the most effective. Bayesian probabilities and credible intervals have different interpretation than classical p-values and confidence intervals. Bayesian results fit well in decision modelling, as resulting posterior distributions can serve as priors in probabilistic cost-effectiveness modelling. Assumptions behind NMA to generate unbiased results were considered valid for IBR vs IDEL+OFA-comparisons, as included patient-populations were nearly identical. Estimates versus PC may be conservative, given higher relative treatments effect in more severe patients. CONCLUSIONS: In absence of direct evidence, NMA-results suggest improved PFS and OS for IBR compared to IDEL+OFA and to PC in R/R CLL-patients with high certainty, and can serve as input in HTA-decision modelling.