OBJECTIVES: Relative survival represents cancer survival in the absence of other causes of death. Cancer Markov models often have a distant metastasis state, a state not directly observed, from which cancer deaths are presumed to occur. The aim of this research is to use a novel approach to calibrate the transition probabilities to and from an unobserved state of a Markov model to fit a relative survival curve. METHODS: We modeled relative survival for newly diagnosed cancer patients through a piecewise Markov model. For each segment we used a constant transition matrix with three cancer states: 1) no evidence of disease, 2) metastatic recurrence and 3) cancer death. We estimated the optimal time points at which the slope of the cumulative hazard changes using a free-knot spline model. We calibrated the transition probabilities using a two-step iterative convex optimization (TICO) algorithm. The dynamics of the disease can be defined as x_t+1= x_tA, where x_t+1 is the state vector that results from the transformation given by the monthly transition matrix A