The curtailment of irrigation on the Klamath Reclamation Project in 2001 is estimated to have cost farmers more than $35 million. This study examines how alternative water allocations among irrigators in the Upper Klamath Basin could have lowered those costs. Per acre marginal water values vary by a factor of 20 due primarily to variations in soil productivity, with the highest productivity lands concentrated in the federal Project. A linear programming model estimates costs for alternative allocations. Findings indicate that compared to the 2001 allocation, costs could be reduced by 75% with a market-based approach.