Two broad surveys of community- level water consumption and pricing behavior are used to answer questions about water demand in a more flexible and dynamic context than is provided in the literature. Central themes of price representation, aggregation, and dynamic adjustment tie together three econometric demand analyses. The centerpiece of each analysis is an exogenous weighted price representation. A model in first-differences is estimated by ordinary least squares using data from a personally-conducted survey of Texas urban water suppliers. Annual price elasticity is found to vary with weather and income, with a value of -0.127 at the data mean. The dynamic model becomes a periodic error correction model when the residuals of 12 static monthly models are inserted into the difference model. Distinct residential, commercial, and industrial variables and historical climatic conditions are added to the integrated model, using new national data. Quantity demanded is found to be periodically integrated with a common stochastic root. Because of this, the structural monthly models must be cointegrated to be consistent, which they appear to be. The error correction coefficient is estimated at -0.187. Demand is found to be seasonal and slow to adjust to shocks, with little or no adjustment in a single year and 90% adjustment taking a decade or more. Residential and commercial demand parameters are found to be indistinguishable. The sources of price endogeneity and historical fixes are reviewed. Ideal properties of a weighted price index are identified. For schedules containing exactly two rates, weighting is equivalent to a distribution function in consumption. This property is exploited to derive empirical weights from the national data, using values from a nonparametric generalization of the structural demand model and a nonparametric cumulative density function. The result is a generalization of the price difference metric to a weighted level-price index. The validity of a uniform weighting is not rejected. The weighted price index is data intensive, but the payoff is increased depth and precision for the economist and accessibility for the practitioner.