In eastern Switzerland there is a substantial weather risk during the wheat harvest. Increasing the number of available combine harvesters reduces harvest losses due to bad weather conditions, whilst increasing machinery costs. Simulating weather risk, we analyse harvest losses and machinery costs. In doing so, we apply both a deterministic and a stochastic approach - the latter by means of a Monte Carlo simulation. The costs per hectare of wheat are higher in the stochastic approach, because bad weather conditions lead to high costs. There is no difference between the two approaches in terms of minimum costs, which in both cases are achieved by a density of 20 harvesters per 1000 hectares. Calculation results are lower than the actual density, which is around 30 harvesters. One explanation for this may be that farmers seem more concerned with decreasing harvest losses than with reducing machinery costs. They are willing to pay more for the harvest process than is strictly necessary, viewing a high density of harvesters as a kind of insurance for weather risk.