Several empirical findings have challenged the traditional trade-off between risk and incentives. By combining risk aversion and limited liability in a standard principal-agent model the empirical puzzle on the positive relationship between risk and incentives can be explained.