This paper investigates to which extent the Dutch feed-in premium scheme for on-shore wind projects has resulted in windfall profits during 2003-2018, a period in which the design of the scheme changed several times. Using Monte Carlo simulations, for 2003, 2009 and 2018, years that represent distinct scheme designs, we estimate the distributions of the required subsidy across virtually all potential on-shore wind projects, and compare them to the granted subsidies. We find that the average windfall profits of randomly drawn projects from the pool of potential investments have decreased over time, largely as a result of differentiating in the subsidy level among projects on the basis of the wind speed at the turbine's location. Despite these improvements, actual investments still experience substantial windfall profits, implying that investors successfully seek out projects that yield the highest windfall profits. Overall, the results imply that accounting for heterogeneity by differentiating in the subsidy level contributes to mitigating windfall profits.