Proponents of the selection effects argument claim that because democratic leaders run a higher risk of losing office than autocratic leaders if they fail to win wars, they are more careful than their authoritarian counterparts in choosing which wars to initiate. The robust marketplace of ideas in democracies also weeds out self-serving or ill-conceived policies and allows democratic leaders to better estimate the chances of victory. Democracies, according to this logic, tend to pick on weak or vulnerable opponents and thus win a disproportionate number of the wars they start. Both quantitative and qualitative evidence, however, challenges this conclusion. The statistical correlation between democracy and victory is not robust to reasonable alternative choices for analyzing the data. In particular, including draws as a war outcome renders statistically insignificant the finding that democratic initiators and targets are more likely to win. In addition, democratic leaders who initiate wars should be optimistic that they will win, but process tracing of the decision by Lyndon Johnson’s administration to escalate the Vietnam War (one of these omitted draws) reveals that top officials knew at the time that escalation promised a costly, protracted stalemate, yet they chose to fight anyway. Moreover, domestic politics, if anything, contributed to Johnson’s decision to fight in Vietnam despite the poor odds of victory because he believed that pulling out would spark a backlash and destroy his Great Society legislative program. The results of this combined analysis raise doubts about the democracy and victory thesis, and should prompt interest in other explanatory variables for military effectiveness.