Normal mammalian somatic cells proliferate a finite number of times in vitro before permanently withdrawing from the cell cycle into a cellular state referred to as senescence. Senescence may be triggered by excessive mitogenic stimulation or by various forms of cellular damage including excessive telomere shortening. Over the past decade, there has been continuing accumulation of evidence that senescence occurs in vivo, that it is relevant to aging and that it has a tumor suppressor function. However, the phenotype of senescence has also been found to include a number of puzzling features, including the secretion of proinflammatory factors that may foster tumorigenesis as well as the senescence of neighboring cells. On the basis of these antagonistic pro- and antitumorigenic effects, and of the observation that many viruses have developed proteins that prevent senescence of the cells they infect, it is argued that the primary function of senescence may have been as an antiviral defense mechanism. Recent progress in understanding how tumor cells evade senescence is also reviewed here.