Markov chain models are commonly used to simulate the natural history of human papillomavirus infection and subsequent cervical lesions with the aim of predicting future benefits of health interventions. Developing and calibrating these models entails making a number of critical decisions that will influence the ability of the model to reflect real conditions and predict future situations. Accuracy of selected inputs and calibration procedures are two of the crucial aspects for model performance and understanding their influence is essential, especially when involves policy decisions. The aim of this work is to assess the health and economic impact on cervical cancer prevention strategies currently under discussion according to the most common methods of model calibration combined with different accuracy degree of initial inputs. Model results show large differences on the goodness of fit and cost-effectiveness outcomes depending on the calibration approach used, and these variations may affect health policy decisions. Our findings strengthen the importance of obtaining good calibrated probability matrices to get reliable health and cost outcomes, and are directly generalizable to any cost-effectiveness analysis based on Markov chain models.