Assessed variation in residential water consumption and evaluated methods of encouraging residents to reduce their consumption. Survey data were collected by mail questionnaire in early 1991, and water consumption figures were recorded between June and August of that year. In Study 1, 3 wks of consumption data were collected from 264 households. A 3-variable regression model (number of residents, clotheswashing machine loads, and property value) accounted for 60% of the variance. Attitudes, habits, and values were very poor predictors of water consumption. In Study 2, 226 households were divided into 3 treatment groups: feedback only, feedback and dissonance, and a control group. A repeated-measures ANOVA revealed that high consumers receiving dissonance and feedback or feedback alone had significantly reduced their water consumption in the treatment period.