Hydrological modeling was used for projecting average annual water availability in Nigeria in the future, comparing a baseline (1976–2005) with a 30-year future (2036–2065) period, simulated under an ensemble of ten climate projections. Simulations converged in projecting by mid-century an increase in water flows for almost half of the country. Models agreed also in projecting decrease and stability in water flows for 13% of the country, while uncertainty covers about one-third of Nigeria. Lack of agreement among different climate models on precipitation and inflow makes it difficult to project how much water will be effectively available in the future for irrigation. Reservoir size is usually designed to ensure sufficient storage providing a given yield based on past climate. The objective of this paper is evaluating what storage investments are suitable for meeting irrigation development targets under as many climate outcomes as possible. A simple methodology for more robust investment planning is suggested and the risk of over- or under-designing storage based on past climate is exemplified. This preliminary study shows that, for more than half of the country, using the historical climate as a guide to the design of future storage might lead to inappropriate investment decisions, resulting in excessive or insufficient capital outlays. The conclusions of the paper do not entail endorsement by the World Bank or its Board of Directors.