We explore how the economically optimal selection of environmental policy instruments is influenced by information available to decision-makers. We also investigate the value of different types of information for environmental management. The focus is on nonpoint source water pollution regulation in the Susquehanna River Basin of Pennsylvania. An extended abstract: Imperfect information about costs and benefits can greatly complicate policy decisions to protect and restore water resources. This has become very apparent to water quality managers in the U.S. in recent years, as they have struggled to comply with the U.S. Environmental Protection Agency's Total Maximum Daily Load regulations. The slow progress has been attributed in large degree to the fact that key information for assessing the condition of streams, lakes, and estuaries, developing sensible plans to restore impaired waters was unavailable and costly to obtain [NRC 2000]. The relationship between agricultural production and damages from water pollution is complex, involving multiple physical and biological links that are not perfectly understood. For example, transport of pollutants off a field to water body depends on stochastic weather events and privately known management practices. The change in water quality, as measured by physical and biological indicators, in response to discharge of agricultural pollutants, is not completely known. Decision-makers have also significant uncertainty about pollution abatement costs, and even more so about economic benefits of water quality improvement. Choices about data collection and analysis to reduce uncertainty and improve water quality programs should be guided by the value of the information for management relative to costs. In this research we examine the value of different types of information, and the effect of information collection on environmental policy performance for the management of agricultural water pollution in the Susquehanna River Basin (SRB) (Pennsylvania). The analysis is based on a model that simulates the effects of water pollution control instruments on polluters' resource allocation decisions, the costs the polluters incur from changes in resource allocation, and the effects of polluters' choices on pollution loads. We group all uncertain factors into three broad categories: economic uncertainty about polluters' abatement costs, hydrologic uncertainty related to the effect of changes in agricultural practices on pollution loads, and damage cost uncertainty, which includes biophysical responses of the water body to pollution and economic valuation of the response. The imperfect knowledge is captured by randomizing the values of model parameters, which affect abatement costs, pollution load resulting from agricultural practices, and the benefits of load reductions. We model five information collection strategies: 1 - no additional information is collected; 2 - 4 - hydrologic, economic or damage uncertainty is resolved; and 5 - perfect information about all uncertain parameters is collected. For each information collection strategy, two types of policy instruments - input taxes and quantity controls - are considered. The expected net benefit maximization is used as criteria for estimating environmental policy performance. The value of information collection is estimated as the expected gain in policy performance due to utilizing information. The results show that performance of price and quantity control differs significantly when economic information is not available, with price mechanisms always outperforming the quantity control. The value of information to the large extent depends on the instrument used in environmental policy. The value of information is greater for quantity controls than for price mechanisms. Even in the case when no additional information is collected, introduction of price control results in significant increase in expected net benefits, while quantity control brings smaller expected net benefit gain. When perfect information is collected, price and quantity instruments perform the same. As a result, the expected increase in net benefits due to information gathering (i.e. the value of information) is smaller for price control than for the quantity control. Information about pollution damage has the highest value for both quantity and price controls. An interesting result is that economic information is more valuable for environmental policy than hydrologic information. Environmental pollution is an economic activity, and addressing the problem requires analysis of environmental policy effects on economic choices (which, in turn, determine pollution loads) and the study of the costs of changes in resource allocation. Our analysis confirms that economic information on both benefits and costs is essential for sound environmental policy design. Reference: National Research Council (NRC). 2000. Assessing the TMDL Approach to Water Quality Management. National Academy Press, Washington, DC.