A gene is said to be polymorphic if more than one allele occupies that gene's locus within a population. In addition to having more than one allele at a specific locus, each allele must also occur in the population at a rate of at least 1% to generally be considered polymorphic.

Gene polymorphisms can occur in any region of the genome. The majority of polymorphisms are silent, meaning they do not alter the function or expression of a gene. Some polymorphisms are visible. For example, in dogs the E locus can have any of five different alleles, known as E, E^m, E^g, E^h, and e. Varying combinations of these alleles contribute to the pigmentation and patterns seen in dog coats.

A polymorphic variant of a gene can lead to the abnormal expression or to the production of an abnormal form of the protein; this abnormality may cause or be associated with disease. For example, a polymorphic variant of the gene encoding the enzyme CYP4A11, in which thymidine replaces cytosine at the gene's nucleotide 8590 position encodes a CYP4A11 protein that substitutes phenylalanine with serine at the protein's amino acid position 434. This variant protein has reduced enzyme activity in metabolizing arachidonic acid to the blood pressure-regulating eicosanoid, 20-hydroxyeicosatetraenoic acid. A study has shown that humans bearing this variant in one or both of their CYP4A11 genes have an increased incidence of hypertension, ischemic stroke, and coronary artery disease.

Most notably, the genes coding for the major histocompatibility complex (MHC) are in fact the most polymorphic genes known. MHC molecules are involved in the immune system and interact with T-cells. There are more than 32,000 different alleles of human MHC class I and II genes, and it has been estimated that there are 200 variants at the HLA-B HLA-DRB1 loci alone.

Some polymorphism may be maintained by balancing selection.

A rule of thumb that is sometimes used is to classify genetic variants that occur below 1% allele frequency as mutations rather than polymorphisms. However, since polymorphisms may occur at low allele frequency, this is not a reliable way to tell new mutations from polymorphisms. A mutation is a change to an inherited genetic sequence.

In the case of silent mutations there isn't a change in fitness, and the pressures responsible for Hardy-Weinberg equilibrium have no impact on the accumulation of silent polymorphisms over time. Most often, a polymorphism is variation in a single nucleotide (SNP), but also can be insertion or deletion of one or more nucleotides, changes in the number of times a short or longer sequence is repeated (both of these are common in parts of DNA that don't directly code for a protein, as are SNPs, but can have major effects on gene expression). Polymorphisms which result in a change in fitness are the grist for the mill of evolution by natural selection. All genetic polymorphisms start out as a mutation, but only if they are germline and are not lethal can they spread into a population. Polymorphisms are classified based on what happens at the level of the individual mutation in the DNA sequence (or RNA sequence in the case of RNA viruses), and what effect the mutation has on the phenotype (i.e. silent or resulting in some change in function or change in fitness). Polymorphisms are also classified based on whether the change is in the sequence of the resulting protein or in the regulation of the expression of the gene, which can occur at sites that are typically upstream and adjacent to the gene, but not always.

Polymorphisms can be identified in the laboratory using a variety of methods. Many methods employ PCR to amplify the sequence of a gene. Once amplified, polymorphisms and mutations in the sequence can be detected by DNA sequencing, either directly or after screening for variation with a method such as single strand conformation polymorphism analysis.