How do varying levels of inter-group contact affect voter preferences in connection with ethnically radical political candidates and parties? Two competing hypotheses have emerged in the last 60 years: the first, known as the group threat hypothesis, argues that voters from an ethnic or religious group in more ethnically or racially heterogeneous districts will exhibit stronger preferences for ethnically radical political candidates. The contact hypothesis argues that groups living in mixed localities are actually less likely to support ethnic radicals. Both perspectives have found empirical support, but no previous study has offered a theoretical explanation for two seemingly contradictory conclusions. We specify just such a theory, arguing that the effect of district level integration is conditioned by the direction of a group’s share of the national population. We test this theory quantitatively using electoral data from Northern Ireland between 1983 and 2010.