Innovative groundwater management strategies are essential to preserve aquifers for crop irrigation. In western Kansas, USA, irrigators self-organized to extend the aquifer’s lifespan by self-imposing groundwater pumping limits enforced over a five-year period. While the five-year groundwater allocation period granted irrigators additional flexibility, it adds a new temporal dimension to their decision-making beyond the typical annual/sub-annual cropping and irrigation decisions. Pumping restrictions, along with uncertain precipitation, complicate multi-year farm planning. We formulated a two-stage stochastic modeling framework to design optimal annual cropping and irrigation allocations under pumping restrictions and uncertain precipitation. Optimal cropping and allocation strategies by the stochastic optimization model significantly outperform observed farmer strategies during the first two five-year LEMA periods (2013-2022) but only outperformed the optimal strategy by the deterministic optimization model assuming long-term average precipitation during drier conditions. We show that optimal cropping decisions shift from predominately corn to sorghum if more stringent pumping restrictions are imposed. Furthermore, irrigators are better off to use less water in the earlier years and saving more water for later years under more stringent five-year pumping restrictions, while they should use more of their allocation earlier under less stringent pumping limits. Extending the duration of the groundwater allocation window allows additional operational flexibility and enhances profits but the marginal gains for each additional year drop off after around seven years. Our versatile modeling framework is applicable to other regions considering groundwater pumping restrictions with the aim of balancing water conservation with farmer profitability and adaptability.