From 1987 to 2008, riskier firms were more likely to be taken over. Yet, on average, the acquirer declined in value by 2.8% when it bought a “risky target” (the third tercile, having an annualized idiosyncratic volatility of 61% or more), but only by 0.6% when it bought a “safe target” (the first tercile, 38% or less). The effect was even stronger for risky targets with positively skewed expected returns. The value difference is robust to controlling for acquirer and target characteristics, and carries over to the joint value change. Riskier target acquisitions also had lower post-acquisition accounting returns. An acquiring-firm CEO fixed effect in the data suggests CEO preferences play a role, which we can trace to several proxies for gambling propensity.