Long term observations of the gene pool of the same and geographically separated populations of Drosophila melanogaster forced us to return to the old idea of De Vries about the existence of mutation periods and fluctuations in the mutation rate with time. A 3- to 5-fold increase of the total mutation rate was estimated by the frequency of lethals, and outbursts of mutability of some X-linked loci were registered. In 1973 the mutation frequency of the singed bristle locus increased in many populations. An outburst of mutability up to 10-2 – 10-3 was accompanied by the unstable conditions of newly arisen mutations, and their ability to produce with a great frequency various allelic variants in the progeny. Two main types of instability have been established: (A) Changes occur in accordance with the ‘all or nothing’ principle: from an extreme mutant expression to a phenotypically normal one, and vice versa; (B.) Mutants with intermediate phenotypic expression give rise to a fan of allelic derivatives in progeny. Mutations occur long before meiosis and proceed at different rates in sexual and somatic cells. The whole body of indirect evidence indicates that the instability is the result of the incorporation of insertion mutations and frequent changes of orientation of some alien segments of DNA. When evaluating the mutation process in natural populations one should take into account that: It is suggested that some infectious agents may be acting synchronously in isolated populations as selective factor and as factor inducing unstable states of genetic loci.