The Author shows that modelling the uncertain cash flow dynamics of an investment project deserves careful attention in real options valuation. Focusing on the case of commodity price uncertainty, a broad empirical study reveals that, contrary to common assumptions, prices are often non-stationary and exhibit non-normally distributed returns. Subsequently, more realistic stochastic volatility, jump diffusion, and Lévy processes are evaluated in the context of a stylised investment project. The valuation results suggest that stochastic process choice can have substantial implications for valuation results and optimal investment rules.