Investment decisions are, as a rule, characterized by uncertainty, irreversibility and flexibility. Simple net present value calculations will not account for these features. In many situations even flexible investment planning with decision trees, which represents the most advanced method of traditional investment appraisal, does not have the capacity to solve practical decision problems adequately. One handicap is a realistic and manageable representation of stochastic variables. It has long been known that stochastic simulation procedures offer a nearly unlimited capacity to represent distributions and stochastic processes. However, a standard simulation will not allow for the consideration of flexibility. The problem is that with a simple forward moving simulation of stochastic paths it is not clear at potential investment dates whether waiting or investing represents the optimal strategy. In this paper we show how stochastic simulation procedures can be integrated successfully into a backward recursive programming approach. The resulting modus operandi can be called "Bounded Recursive Stochastic Simulation" (BRSS). We use this efficient combination of simulation and dynamic programming to answer the question whether farmers should buy sales contracts which guarantee fixed prices for rye in the future. The results of the model affirm the importance of uncertainty and flexibility for investment decisions. They also show that the actual conditions offered by the wholesale buyer are not economically attractive for farmers, unless they are extremely risk averse. Thus, model results coincide with the empirical evidence that farmers do not enter these contracts.