Many applied studies ask people to rate reasons why they perform a behavior. The assumption is that strongly endorsed reasons are motives underlying behavior. However, this assumption rarely is examined empirically. Moreover, the self-reported reasons literature has lacked theoretical attention. In this study, we presented a formal model of self-reported reasons, entitled reasons theory (RT). RT formally conditions reason type (i.e., reasons for and reasons against) on behavioral frequency/intention. RT was compared to a standard reasons approach that does not present formal reason postulates. Results showed that RT explained variance in behavior over and above that explained by standard reasons, and that correlational coefficients from RT better matched mean reason results than did standard reasons. RT results were also more substantively interpretable than were standard reasons. Lastly, several boundary conditions were discussed, including the relationship between reasons and emotions.