Explicit consideration of benefits and costs is needed to determine how much to spendfor public data. It is argued that benefit-cost analysis is also feasible. In support of feasibility, a benefit-cost analysis of the 1970 census is discussed in some detail, with emphasis on allocative uses of the data. Other precedents are discussed for evaluating the production, analysis, and dissemination of forecasts and projections, data used for determining allocations, and physical monitoring data. The greatest potential value of benefit-cost analysis of data may be for social monitoring data, such as social and economic indicators.