We present a bio-economic model by combining a process-based grassland simulation model with an economic decision model that accounts for income risks and yield quality. The model is used to examine optimal nitrogen (N) application rates in a grass–clover system in Switzerland under current and future climatic conditions. Results for present-day climatic conditions suggest that an increase in N inputs has positive effects on yields but also leads to higher yield variability, yield distributions more skewed to the left and therefore higher downside risks. As a result, accounting for farmers’ risk aversion in solving the optimisation problem leads to lower optimal N inputs. Simulations with a climate change scenario that predicts higher temperatures throughout the year and lower rainfall amounts during the growing season indicate higher yields, increasing yield variability, and changes in yield quality. By allowing herbage prices to vary as a function of yield quality, we find overall lower optimal N inputs and more marked effects of risk aversion on optimal N levels under climate change than under present conditions. However, disregarding yield quality in solving the optimisation problem gives higher optimal N inputs under future conditions.