OBJECTIVES: The choice of a modeling approach is guided by key criteria including time, interaction between individuals, and the unit of analysis (cohort or individuals level models). Despite Markov cohort modeling (MCM) is widely used in the literature, the use of partitioned survival (PS) models tends to increase. Our objective is to explore the rationale for selecting either a Markov modeling approach or a PS approach to carry-out a cost-effectiveness analysis in oncology. Our study focuses on the differences between the two approaches. METHODS: A literature review focusing on survival modeling in economic evaluation was performed in order to establish a list of differences between the two modeling approaches. Besides, we reviewed NICE's technology appraisals (TA) in oncology medicines over the last two years (2013-2015) to analyze the practices and the arguments put forward to justify modeling choices. Data collected for each TA included: model type, rationale for model selection, health states, hypotheses, survival analysis, clinical data sources and the treatment of uncertainty.   RESULTS: Twelve economic evaluations in oncology were submitted to NICE by pharmaceuticals companies (PC) between 2013 and 2015. Seven PC submitted a MCM, two a PSM, two a semi-markov partitioned survival model, and one a semi-markov model (SMM). Differences between modeling techniques were classified into four items: clinical data sources (e.g. published aggregated data for MCM and limited IPD for PSM), structure (e.g calculation of transition probabilities for MCM), hypotheses (e.g. same transition probability of death between two health states for MCM), flexibility of the model (e.g. access to patient level data for comparators required in PSM). CONCLUSIONS: Being a more flexible modeling technique, Markov models remain more frequently used compared to PSM. Nevertheless, PSM represent a more straightforward option when patient level data are available but are inappropriate when such data are not accessible for comparators.