When cells are induced to undergo apoptosis in the presence of general caspase inhibitors and then returned to their normal growth environment, there follows an extended period of life during which the entire cohort of mitochondria (including mitochondrial DNA) disappears from the cells. This phenomenon is widespread; it occurs in NGF-deprived sympathetic neurons, in NGF-maintained neurons treated with cytosine arabinoside, and in diverse cell lines treated with staurosporine, including HeLa, CHO, 3T3 and Rat 1 cells. Mitochondrial removal is highly selective since the structure of all other organelles remains unperturbed. Since Bcl2 overexpression blocks the removal of mitochondria without preventing death-inducing signals, it appears that the mitochondria are responsible for initiating their own demise. Degradation of mitochondria is not in itself a rare event. It occurs in large part by autophagy during normal cell house-keeping, during ecdysis in insects, as well as after induction of apoptosis. However, the complete and selective removal of an entire cohort of mitochondria in otherwise living mammalian cells has not been described previously. These findings raise several questions. What are the mechanisms which remove mitochondria in such a 'clean' fashion? What are the signals that target mitochondria for such selective degradation? How are cells that have lost their mitochondria different from rho0 cells (which retain mitochondria but lack mitochondrial DNA, and cannot carry out oxidative phosphorylation)? Are the cells which have lost mitochondria absolutely committed to die or might they be repaired by mitochondrial therapy? The answers will be especially relevant when considering treatment of diseases affecting long-lived and non-renewable organs such as the nervous system.