When should we expect an opposition group to select an extremist leader or representative? This paper shows the important role of corruption for this choice. Moreover, we show an important asymmetry in the role of corruption, in that the effect on extremism exists only within the opposition group. When the elite has greater ability to use corruption to obtain a better bargaining outcome from the opposition group leader (political corruption), then the equilibrium selection of group leader is more likely to be extreme. On the other hand, the perception of an existing rent extraction by the elite in power may determine the opposite effect within the majority group. We provide strong evidence for these novel predictions using the random audits data in Brazil as exogenous corruption signals, verifying that only within the opposition (to state-level incumbents) the signals determined an extremism drift in voting. Finally, we extend the analysis to extremism and conflict risk in divided countries.