The competition or rivalry for the use of water resources among economic sectors of the Pacific Northwest and among geographical regions of the western United States has intensified in recent years. This rivalry and the long run prospects for water shortages have increased the demand for research concerning the productivity of this resource in alternative uses. This demand exists because the distribution and use of water resources require investment which typically comes from both public and private sources. Private and public planning groups seek answers to questions regarding future water resource development alternatives. Agriculture has historically been a major user of water in the Pacific Northwest. A substantial portion of total investment in water resource development has also been in agriculture. As a result water use planners and decision making bodies are necessarily interested in water use in agriculture. The success of water resource planning requires answers to questions regarding the value of the productivity of water in all its major uses, including various aspects of water use in agriculture. Different aspects of water use in agriculture which are important to decision makers include (1) the value productivity of various kinds or types of water resource investments, (2) the value productivity of water in various kinds of agricultural production in different geographical areas, and (3) the returns to private and public investment in agricultural water resources. This study was directed to providing answers to these questions. Pacific Northwest agriculture was studied from this viewpoint. Agricultural water resources were classified as irrigation, drainage, and water related Agricultural Conservation Program (ACP) practices. These are the major classifications of water resources in which investments are made in the Pacific Northwest. Production function analysis was selected as a method of investigation. Production functions were estimated for five areas or subregions in the Pacific Northwest. These areas are composed of counties with similar patterns of production. The Agricultural Census was the primary data source, supplemented by related. U.S. Department of Agriculture publications, and various state publications. Ordinary least-squares regression (OLS) techniques were employed to derive the initial estimates of the parameters of the production function models. Tests for detecting interdependence within the independent variable set of the models revealed a considerable degree of instability in the OLS parameter estimates. This condition makes the OLS solutions (and various derivations) particularly vulnerable to change from measurement error, poor model specification, and equation form. A prior information model was selected to explicitly include available prior knowledge in the estimation process. The model selected allows (1) tests of comparability of the two information sources (prior and sample), (2) over-all contribution of prior information to the new solution set, and (3) derivation of percentage contribution of the two information sources to individual parameter estimates. The results of the study indicate that no reliable estimates of value of production from drainage and ACP were possible from the sample information. Returns to irrigation were considered lower than expected in two of the farming areas and higher than expected in another. Estimated returns were high in the area which produces primarily field crops (about nine dollars per acre foot). The area has a small level of current irrigation development. Indications are that irrigation development is probably beyond the optimum level in the area where most large projects have been developed in the past (less than four dollars per acre foot). Future development would be most profitable (assuming equal development cost) in the dryland field crop area. Estimated returns to other factor inputs indicate (1) low returns to labor in two areas, (2) generally high returns to current operating expenditures, and (3) low returns to machinery capital. Returns to cropland were about as expected in two areas (five to seven percent) but low in two other areas (about two percent). Indications are that labor mobility should be increased in the area and that future land development should be in the livestock-field crop and the field crop areas rather than the coastal area or the west-central valley areas (primarily the Willamette Valley).