This paper introduces the method of importance-performance analysis of citizen surveys, a useful approach to understanding citizen satisfaction with local government services. Using data from a US national online panel, we directly compare two approaches to importance-performance analysis: one employing an explicitly stated measure of importance, the other using a measure of importance derived from regression analysis. The different results that the two approaches give suggest that local government administrators and policy analysts arrive at distinctly different conclusions depending on which importance measure they use. These differences are illustrated by simulating the change in citizen satisfaction that would result from improvement in the top-rated services according to each measure. Research and policy implications are discussed.