The following article illuminates existing challenges and restrictions when implementing available stochastic user behavior models in building performance simulation (BPS). 24 occupancy behavior models from the literature containing a clear mathematical description are attempted to be coupled with a BPS model in a case study. Different methods, amongst others co-simulation approaches benefitting from the Functional Mock-up Interface (FMI) standard, were investigated to realize the implementation. The majority of OB models were coupled successfully with the BPS; however, some were not. The reason for the failed coupling is rather based on the restriction of OB models for BPS use than the coupling methods. Generally, OB models are based on stochastic modeling, while BPS requires a clear decision, a trigger for further interaction. Some OB models do not provide an output in such a binary form. Therefore, it is difficult to use these models in BPS without any assumption from the modeler. Furthermore, the majority of OB models lead to a state change depending on a comparison between its computed probability and a random number, which conflicts with the reproducibility of BPS results. In addition, some OB models result in an improper behavior without a reversal function or hysteresis. Based on the case study, these issues and requirements for OB models for the use in BPS as well as the advantages and disadvantages of various coupling approaches with BPS are discussed.