Abstract: Much recent conservation effort has focused on genetic events in small populations, such as threatened or endangered species on the verge of extinction. However, the overwhelming causes of population reductions and extinctions worldwide are habitat destruction and the introduction of exotic species of parasites, predators, and competitors. The restoration and maintenance of healthy habitats and ecosystems should be of great concern to a mature science of conservation biology. The long-term preservation of biodiversity requires understanding not only the demography and genetics of small populations but also the ecology and evolution of abundant species. Here we show that in constant or unpredictable environments genetic variance reduces population mean fitness and increases the risk of extinction. In predictable, highly variable environments genetic variance may be essential for adaptive evolution and population persistence. Most of the characters of interest to ecologists and evolutionary biologists are quantitative characters influenced by many genes and environmental factors. Meristic and threshold characters also are amenable to analysis using quantitative genetic methods (Wright 1968, ch. 15; Falconer 1989). As examination of the fossil record attests, quantitative characters are of great importance in adaptive evolution (Simpson 1953; Carroll 1988). Although adaptive evolution can occur by mutations of large effect, the divergence in the quantitative traits that distinguish both different populations within a species and closely related species usually has a polygenic basis (Wright 1968, ch. 15; Lande 1981; Coyne 1985). No comprehensive evaluation of the importance of genetic variability in quantitative traits to population persistence and adaptation exists currently. In the short-term, genetic variability is often less critical than other determinants of population persistence (Lande 1988), but in the long-term, it can play the decisive role in allowing a population to persist and adapt in a changing environment. The rate of evolution in the mean phenotype in response to selection on a single quantitative character is proportional to the product of the additive genetic variance in the character and the intensity of directional selection (Lande 1976; Falconer 1989). However, genetic variability is thought not to be the rate-limiting factor in long-term evolution. Instead, long-term rates of evolution and adaptive radiation are constrained by ecological opportunity (Simpson 1953, pp. 77-80; Wright 1968, p. 520). That the shortand the long-term views are not inconsistent can be seen in a model of the common situation in which natural selection acting on a quantitative character (other than fitness itself) favors an intermediate phenotype. In this situation the rate of evolution in the character is limited not only by the magnitude of the additive genetic variance in the character but also by the rate of change in the optimum phenotype as the environment changes. An intermediate-optimum model such as that which follows also demonstrates that genetic variability may be either beneficial or detrimental, depending on the pattern of environmental change.