We revisit the incentive effects of elimination tournaments with a fresh approach to identification, the results of which strongly support that performance improves under the threat of elimination and does so, but only in part, due to increases in risk taking. Where we can separately identify changes in risk-independent performance and risk taking, our estimates suggest that 23 percent of the improvement in performance induced by potential elimination is due to productive increases in risk taking. These effects are concentrated among those closest to the margin of elimination and among lower-ability competitors.