Genetic drift is a mechanism of evolution that biologists believe to occur in concert with natural selection. It is a statistical effect that arises due to the influence of chance on the survival and reproduction of individuals. The effect is to cause certain traits, and the alleles that confer them, to become more common or more rare over successive generations.
In statistical terms, drift is a so-called sampling phenomenon. Many sources of mortality, such as diseases for which there is no immunity, may be regarded as randomly "sampling" a population or species. That is, they select some proportion of individuals at random for death. Sexual selection by a particularly fertile mate, which tends to make a trait or allele more common, also may be seen as random sampling. Because selection is random, alleles on average will be selected in proportion to how common they are. But because the sample size, the population size and the "frequencies" of alleles are finite, deviations from the average or mean are prone to occur. The frequency of an allele drifts to the extent that the deviations up or down over successive generations do not exactly balance out.
Due to the law of large numbers, these deviations are proportionately larger in small populations and rare alleles. When many individuals carry a particular allele, and all face equal odds, the number of offspring they collectively produce will rarely differ from the expected value, which is the expected average per individual times the number of individuals. But with a small number of individuals, a lucky break for one or two causes a disproportionately greater deviation from the expected result. Small populations therefore drift more rapidly than large ones. This is the basis for the founder's effect, a proposed mechanism of speciation.
If the number of individuals who carry an allele drifts to zero, so that no individuals are left to reproduce it, the allele disappears forever. Similarly, if all but one allele disappears, the proportion of individuals who carry it will never stray from 100%. That is, until in at least one individual a spontaneous mutation or other genetic change affects that carrier's allele.
The inevitability of drift poses biologists with a problem. How does one know whether a characteristic that all the individuals of a species possess--yellow spots, for example--represents an adaptation to selective conditions, or represents an accident? It can be difficult to say, and indeed to a degree the answer may be "both." Teasing apart the relative influence that drift and selection have had over the course of evolution, both with respect to individual species as well as with regard to the history of life in general, is an important task of evolutionary biology.