The term covers a potpourri of changes. The quintessential "point mutations" include the substitution of one nucleotide pair subunit of a DNA molecule for another (eg cytosine-guanine for a thymine-adenine pair), and the deletion or addition of one or several base pairs in a gene. At the other end of the spectrum are gross rearrangements in a chromosome, such as the inversion of an internal segment, and the losses of part or all of a chromosome. These are called "chromosomal mutations" instead of gene mutations.

Various other types of genetic change are not considered to be mutational because they are judged too unstable, deliberate, frequent or functional to qualify. "Mutation" is generally envisaged to be a permanent, extremely rare, random, unpredictable, spontaneous error in inheritance. Therefore, various classes of unstable or highly revertible genetic changes in plants are called paramutations, genotrophs or somaclonal variants instead of mutants. Correspondingly, mammals show a variety of "transgenerational carryover effects", and organisms from bacterial to humans have special gene processing enzymes to deliberately alter genes in particular ways. None of these sorts of changes is considered to be true mutation. The point of importance is that a core of gene alterations are unambiguous mutations, with all of the defining characteristics listed above. However, others grade into gene correction, infection, physiological adaptation, differentiation, unequal crossing over, differential replication, genetic anticipation, gene imprinting and so on.

Some changes in chromosomes or genes are carried out by enzymes so deliberately, precisely and usefully that they must be viewed as natural genetic engineering instead of "mutation". Organisms do so only very sparingly. On one side of the coin, its limited occurrence has been taken to imply the insignificance of natural genetic engineering. The flip side is that where it does occur it is important. Humans could not survive if their immune cells did not cut and splice their genes for immune proteins to a degree unmatched by the boldest ambitions of any genetic engineering laboratory. Frog's eggs would be inviable, some trypanosomes would loose the keystone of their pathogenicity and the whole genetic strategy of ciliates would fail without cells manipulating the structure of their genes in the same sorts of ways - and even by using the some of the same enzymes - that genetic engineers do. All but the simplest sorts of point mutations are brought about or tailored by enzymes. Evolutionists, geneticists and pathologists are just beginning to realize the role of enzymatically produced mutation in cells .

Contrary to some popular belief, most mutations cause notably small changes in the phenotype of the organism. Of course a mutation can distort an organism so profoundly that it dies, even at an early embryonic stage, but these are rare. The vast bulk of the DNA in mammals appears to have no coding function what so ever and has been called "junk DNA". Mutations in it typically are "silent" without any effect on phenotype. Even many point mutations within a gene are totally silent due to the way genes code for protein sequences. And mutations which do alter the amino acid sequence of an enzyme still may not detectably change its functional competence. One further unexpected discovery from genetic engineering experiments is that any of a substantial number of genes of mice and other laboratory species can be completely destroyed with no, or only very minor, detectable effects on the phenotype.