A model of demand for potable water is developed and estimated simultaneously with observations on rural water district customers. The estimation technique incorporates the error structure in an attempt to provide more efficient estimates and eliminate potential ordinary least squares bias in single equation models of demand for goods sold through block rate pricing systems because of possible price endogeneity. The model also tests alternative specifications for price variables. The simultaneous equation model estimates from three‐stage‐least squares were slightly more efficient compared to two‐stage‐least squares estimates. The findings are not generally supportive of the Taylor‐Nordin specification of consumer demand for block rate priced potable water. The similarity between three‐stage‐least squares and single equation ordinary least squares estimates provides some validation for the use of simpler single equation demand models of potable water when water is sold under block rate pricing schedules.