Abstract The use of a single electricity price has been seen as a major shortcoming of econometric models of residential electricity demand. It has been suggested that demand estimation should be based on the full rate schedule. This suggestion is evaluated here by examining demand estimates for 27 investor-owned US utilities over the period 1957–1972. The specification errors resulting from using an average revenue price are measured. It is found that statistically valid demand estimated do not necessarily require information on tariff schedules, but that state-wide aggregates of elasticity may often be inapplicable to individual utilities.