Electoral participation in the United States is examined to provide a clearer account of the effect of the registration requirement on individual voting behavior. Pooling NES data from 1980, 1984, and 1988, I first model, with traditional and selection bias techniques, the full electorate to distinguish among three groups: nonregistrants, registered nonvoters, and voters. Analyses limited to recent movers then reported to understand more fully the forces associated with the actual decision calculi of registering and voting. The influences of many factors commonly accepted as important determinants of voting are disentangled, and their effect at each stage is ascertained. Factors yielding inconsistent effects in previ research or believed to be unimportant—such as race, gender, attitudes toward the candidates, and trust government—are shown to deserve closer scrutiny by electoral scholars.