The paper proposes a simulation-based approach to forecasting of a stochastic demand to a supply chain. The explored forecasting method is based on information about a provided service level that is estimated by simulating processes in the considered supply chain. A structure of an integrated system that incorporates a simulation model and a forecasting model is described. The simulation model is used for evaluation of a current service level, provided by the supply chain. The forecasting model transfers information about the achieved service level into a forecasted value of the end customer demand. The runtime simulation technique is discussed as a method to model, during a simulation run, the end customer demand dependency on the achieved service level