This paper studies the interplay between violence and electoral politics. Often political parties not only compete electorally with each other, but also use violence in the electoral processes. This paper game-theoretically models the interplay of such ‘extra-electoral’ investments and electoral outcomes in an otherwise standard probabilistic voting model. We find that the political party that is likely to be more popular is also more likely to expend greater resources towards ‘extra-electoral’ elements, in turn spawning greater violence, even when such investments are disliked by all voters. We also look at some plausible extensions of the benchmark model where the basic conclusions still hold true. The essential flavour and predictions of the model are borne out by several historical and contemporary instances of politics.