Background: Cost-utility analyses of later-line treatments for metastatic colorectal cancer (mCRC) differ based on network meta-analysis (NMA) approach: conventional constant hazard ratio vs novel survival curves methodologies. Methods: We conducted a CUA estimating the incremental cost utility ratio (ICUR) over a 5-year time horizon from the US payer perspective. The 3-state partitioned survival models included efficacy inputs from a novel one-step multivariate NMA of progression-free and overall survival curves and from a conventional constant hazard ratio NMA. Utilities, as well as drug acquisition, administration, adverse event, monitoring and end of life costs were sourced from literature. We compared results and examined statistical factors in either approach that may explain the observed results. Results: The multivariate NMA produced lower estimates of survival benefits over placebo compared to the NMA based on constant hazard ratios, at the same incremental costs (Table). This reduction in incremental benefits led to higher estimates of cost-effectiveness and cost-utility when using the multivariate NMA. Probabilistic sensitivity analyses showed greater uncertainty in the model based on the constant hazard NMA. Conclusions: The multivariate NMA produced lower estimates of survival benefits over placebo compared to the NMA based on constant hazard ratios, at the same incremental costs (Table). This reduction in incremental benefits led to higher estimates of cost-effectiveness and cost-utility when using the multivariate NMA. Probabilistic sensitivity analyses showed greater uncertainty in the model based on the constant hazard NMA.