Evolutionary considerations predict that rate of aging should vary in direct relation to the mortality rate of presenescent young adults (extrinsic mortality rate) independently of differences in physiology, such as rate of metabolism. This prediction emerges from theory irrespective of the particular genetic mechanisms responsible for variation in aging. Yet this critical relationship has not been confirmed in comparative studies of natural populations. In the present analysis, rate of aging is estimated by the rate of increase in mortality rate (mx) with age (x). Comparisons between natural and captive populations of birds suggest that the Weibull model (mx = m0 + alphaxbeta) provides a better description of aging than the Gompertz model (mx = m0esigmax). Rate of aging is quantified by the parameter omega (dimension: 1/time), which is calculated from the Weibull parameters alpha and beta (omega = alpha1/(beta+1)). In this analysis, rate of aging in birds and mammals is directly related to extrinsic mortality (estimated by the initial mortality rate, m0) independently of taxonomic group and of variation in body size and, by implication, metabolic rate. When time is expressed in years, rate of senescence is related to initial mortality rate by omega = 0.294m0(0.367). This result implies that natural selection in response to variation among taxa in m0 has resulted in the evolutionary modification of factors that influence the rate of aging in natural populations. The potential strength of selection on factors that could further reduce rate of aging is indicated by the proportion of deaths due to aging-related causes. Although species with low initial mortality rates also exhibit reduced rates of increase in mortality rate with age (i.e., delayed senescence), the relatively high proportion of aging-related deaths in such species suggests that further evolutionary responses leading to long life are severely constrained. This argues against mutation accumulation and antagonistic pleiotropy as genetic mechanisms underlying senescence and suggests, instead, that rate of aging represents a balance between wear and tear, on the one hand, and genetically controlled mechanisms of prevention and repair, on the other. Evidently, remedies for extreme physiological deterioration in old age either are not within the range of genetic variation or are too costly to be favored by selection.