This paper re-examines the relationship between firm performance and CEO turnover. We do away with the distinction between forced and voluntary turnovers and introduce the concept of performance-induced turnover, defined as turnover that would not have occurred had performance been better. We show that more than 40% of all CEO turnovers are performance induced, and more than 50% of turnovers in the first eight tenure years. This far exceeds the frequency of forced turnovers identified in prior studies. We also find that the effects of performance on turnover are as high in the first five tenure years as in the next five, and that the effects decline only after tenure year 10. Further, CEO departures in all tenure years respond strongly to recent performance but are almost insensitive to performance in the more distant past. These results reject the standard model of CEO turnover in which boards learn from firm performance about constant CEO ability.