Reports a longitudinal analysis of residential energy conservation by 519 residents in a medium-size US metropolitan community. Mail panel surveys were conducted during winter months of 1979, 1980, and 1981. Results are presented in the form of a multivariate causal model with cross-lagged correlations over time. Perceptual, attitudinal, and behavioral variables were found to be the major causal factors, with certain other variables having secondary effects. Demographic variables were not significant in explaining actual conservation. The model presented is a rotational, parsimonious one that suggests several avenues for public policy, including indications of potentially effective conservation messages, audience segmentation, and time required for such interventionist strategies to show results.