Since the discovery that the cysteine protease CED-3 was essential for developmental death in the nematode C. elegans, the search has been on to identify homologous proteases governing mammalian apoptosis. Fourteen of these proteases, now called caspases, have been found to date, and studies with natural or chemical inhibitors, and more recently knock-out mice, confirmed the involvement of at least a subset of these proteases in various forms of mammalian apoptosis. However, there has been recent evidence that some apoptotic morphologies, such as cell shrinkage, membrane blebbing and nuclear condensation, are not blocked by caspase inhibitors and that the cells continue to die in a protracted and inefficient manner. This has led to the notion that caspases are not required for all aspects of apoptosis in mammals. Here we review the current knowledge about caspase-independent apoptosis, discuss the strengths and weaknesses of the reasoning that led to its proposition and provide insights into its possible regulation and physiological significance.