The social learning theory of policy decisionmaking provides clear expectations for the temporal diffusion of policy through the American states (Gray, 1973; Rogers, 1995; Walker, 1969). But the diffusion of morality policy, with its technical simplicity, potential for high salience, and debate over basic moral values, may be driven by nonroutine decisionmaking. We hypothesize that morality policy will diffuse in patterns that depend on the purity of the morality debate surrounding it and the distribution of citizen values regarding it. We identify three distinctive temporal diffusion patterns for state death penalty policies that vary on these characteristics. We conclude that political decisionmaking driving some morality policy diffusion involves less policy learning, and more concern about majority public opinion, than does nonmorality policy.