OBJECTIVES: The memory-less feature of Markov models can be a limiting factor when treatment-sequencing needs to be modeled and the transition probability in second- and subsequent-line treatments are not constant. Although tunnel-states are commonly used to model time-dependency, they can become unruly. Hence, patient simulation models and/ or sophisticated software packages such as R are required to model complex time dependency. An alternate method of using nested markov models was presented at a previous conference to model time-dependency in treatment sequence for a hypothetical model in Excel. This method is now applied to a published model in epilepsy and results are validated using real data. METHODS: The Wilby 2004 epilepsy model is used as a reference to derive model inputs and validate results. It is a probabilistic treatment sequencing decision model in epilepsy implemented using R. The nested markov method involves first disaggregating the model by treatment, then combining the net present values of each treatment into the treatment sequence by weighting proportional to the time spent in the sequence, lastly followed by further discounting to account for placement in the sequence.  Results obtained using the nested markov methods are validated with those published in Wilby 2004.  RESULTS: Quality-adjusted life-years obtained with the nested Markov modeling approach were similar and were within the confidence intervals of results obtained by Wilby 2004. CONCLUSIONS: Nested markov models can be a simple alternative to model time-dependency if transparency and less intense computational programming are required. It represents a straightforward and intuitive approach to modeling a fixed treatment sequence, however, it may not be suitable if the position in a sequence is inter-changeable, and treatment effectiveness depends on the position in a sequence (e.g. cancer therapies where disease progression impacts treatment effectiveness).