OBJECTIVES: Breast cancer is the most common malignant disease in Western women. In the ONCOTYROL research center, a decision-analytic Breast Cancer Outcomes & Policy (BCOP) model is being developed to evaluate the cost-effectiveness of the new 21-gene assay that supports personalized decisions on adjuvant chemotherapy. Model validation is essential to build confidence in the model results and to influence decision makers. Based on the new ISPOR-SMDM best practice recommendations, the process of model validation will be presented. METHODS:   The 21-gene assay was evaluated by simulating a hypothetical cohort of 50year old women over a lifetime time horizon, adopting a societal perspective. Main model outcomes were life-years gained, quality-adjusted life-years (QALYs) gained and costs. The major focus of the presentation is on cross validation, i.e. the comparison of modeling results between the discrete event simulation (DES) BCOP-model and the Markov model of the THETA (Toronto Health Economics and Technology Assessment) Collaborative. Therefore, the BCOP-model has been populated with the Canadian parameters of the THETA-model. RESULTS: Cross validation started with comparison of model parameters related to the natural history of the disease (undiscounted life years, number of breast cancer recurrences/deaths). Thereafter, quality of life and cost outcomes were compared. The comparison included point estimates of the outcomes of the deterministic analysis of the Markov model as well as the probabilistic run with the DES results and combination (ICERs). The absolute differences of expected life years gained for women after surgery ranged from -0.35 to 0.43 years depending on the treatment strategy for specific risk groups. For the probabilistic analysis, confidence intervals as well as distributions of model outcomes were compared. CONCLUSIONS Cross model validation is a suitable approach to identify and correct modeling errors and to explain remaining differences of modeling results.