Using 2,956 CEO turnovers from 1993 to 2009, I find that default probability is useful in understanding and predicting forced CEO turnovers for non-distressed firms, controlling for conventional performance measures, such as stock performance. The high predictive power is not explained by...