Senescence is the post-maturation decline in survivorship and fecundity that accompanies advancing age. Two main evolutionary theories have been proposed to account for senescence. (1) The mutation-accumulation theory. Deleterious mutations exerting their effects only late in life would tend to accumulate, because of their minimal effects on fitness. More precisely, exclusively late-acting deleterious mutations will attain higher equilibrium frequencies under mutation--selection balance than will mutations that act early, resulting in lower mean values for fitness components late in life (ref. 3, p. 218). Medawar emphasized the possibility that this effect would be enhanced by selection of modifiers that postpone the age of onset of genetic diseases. (2) The pleiotropy theory. Williams suggested that many of the genes with beneficial effects on early fitness components have pleiotropic deleterious effects on late fitness components, but are nevertheless favoured by natural selection. (These theories are based on the decline with age in the effect of age-specific fitness-component changes on total fitness (ref. 3, pp. 206--214 and refs 4, 5). Either or both of these theories could apply in any particular population.) Selection experiments in Drosophila and Tribolium support the pleiotropy theory, although one such experiment gave results that only bordered on significance, but the mutation--accumulation theory has never been tested. The present results provide evidence for the pleiotropy theory, but do not support the mutation--accumulation theory.