We estimate whether risk preferences are affected by traumatic events by using a unique survey of Sri Lankan twins which contains information on individual's exposure to the 2004 Indian Ocean Tsunami, participation as a combatant in the civil war, validated measures of mental health and risk preferences, and a rich set of control variables. Our estimation strategy utilises variation in experiences within twin pairs and allows us to explore whether preference changes are driven by wealth shocks and/or changes in mental health. We find that both events lead to less risk aversion, a result that is not driven by mental health or wealth changes.