OBJECTIVES: Published cost-effectiveness analysis (CEA) models, especially those which have been submitted to health technology assessment (HTA) authorities, are valuable in early phase CEA of interventions in the same or similar disease areas. However, even well-documented Markov models can be difficult to replicate. Additionally, differences in disease states, treatment effects, and patient populations can render a successfully-replicated Markov model non-informative for early phase investigations. As a basis for future adaptation and expansion, a published chronic migraine CEA Markov model was replicated in TreeAge Pro 2014 as both a Markov model and as an individual-based state-transition (Monte Carlo microsimulation [MCm]) model.  METHODS: The published and replicated Markov model results were compared for both base case and sensitivity analyses. Patient subgroup Markov transition probability (MTP) matrices were implemented in the MCm, with assumptions regarding unpublished information on post-initial cycle state transitions. These assumptions involved subgroup treatment effects, patient decision discontinuation, and treatment stopping rules. The overall patient population (OPP) MTP matrices generated by the MCm were loaded into the Markov model to assess the validity of the assumptions.  RESULTS: Incremental costs and quality-adjusted life-years (QALYs) between intervention with onabotulinumtoxinA and placebo were produced. Differences between published and replicated Markov model incremental cost and QALY results were small for the base case (0.0%, -1.1%) and selected sensitivity analyses (maximum differences [1.9%, -1.9%]). Base case differences between published and the replicated Markov model (using MCm-derived OPP MTP matrices) incremental cost and QALY results were larger but acceptable (5.2%, 11.7%). This reflects the assumptions involved with the approximation of the patient subgroup MTP matrices.  CONCLUSIONS: The study demonstrates how a published model can be replicated and adapted for early phase CEA investigations, allowing for modeling of OPPs, treatment effects, treatment discontinuation, etc. that differ from the published model.