This paper develops a multi-stage stochastic programming (SP) approach aiding a European company in currency hedging. While cash management concerns several major currencies, our pilot model deals with US$ and € only. Equilibrium correction models, Taylor rule based models and a random walk model are compared for exchange rate prediction. Risks related to exchange rate and sales forecast errors are hedged. Numerical results indicate that the current hedging policy roughly amounts to the same as no hedging at all. We demonstrate how repeated hedging activity reduces risk and thereby suggests avoid excessive risk averse behavior. Out-of-sample tests over the period 2004–2013 indicate that optimized hedging can increase net profits before taxes by about 20% over current policy. Average performance improvement of the random walk model is outperformed in terms of profit improvement by all other models we considered. Out-of-sample results show that single-stage SP yields approximately the same average improvement as multi-stage SP but the latter is more robust in terms of reduced variance.