Methods available to measure the performance of public investment are developed and tested with specific attention to water resources projects for flood control, municipal and industrial water supply, and recreation. The direct measurements attempt to measure variables showing the net benefits from each project purpose while the indirect techniques measure the growth of surrogate variables which reflect the impact of an entire project. The research uses these methods to measure the ex post performance of two small watershed projects in Georgia, Rooty Creek and the Little Tallapoosa River. By observing the growth of income in analogue undeveloped watersheds the discrepancies between ex ante and ex post estimates of costs and benefits from these projects are discussed, and suggestions are made on ways to improve future agency benefit-cost analyses. (Modified author abstract)