I examine the impact of risk preferences on efforts and winning probabilities in generalized Tullock contests between two players. The theoretical analysis yields two ma in results. First, in symmetric contests, a higher common level of the agents ́ risk aversion will lead to lower aggregate effort if it comes along with an increase in downside risk aversion. Second, in many asymmetric contests, higher risk-aversion will be a disadvantage if the agent is also more downside risk-averse