In genetics, an isochore is a large region of genomic DNA (greater than 300 kilobases) with a high degree of uniformity in GC content; that is, guanine (G) and cytosine (C) bases. The distribution of bases within a genome is non-random: different regions of the genome have different amounts of G-C base pairs, such that regions can be classified and identified by the proportion of G-C base pairs they contain.

Bernardi and colleagues first noticed the compositional non-uniformity of vertebrate genomes using thermal melting and density gradient centrifugation. The DNA fragments extracted by the gradient centrifugation were later termed "isochores", which was subsequently defined as "very long (much greater than 200 KB) DNA segments" that "are fairly homogeneous in base composition and belong to a small number of major classes distinguished by differences in guanine-cytosine (GC) content". Subsequently, the isochores "grew" and were claimed to be ">300 kb in size." The theory proposed that the isochore composition of genomes varies markedly between "warm-blooded" (homeotherm) vertebrates and "cold-blooded" (poikilotherm) vertebrates and later became known as the isochore theory.

The isochore theory purported that the genome of "warm-blooded" vertebrates (mammals and birds) are mosaics of long isochoric regions of alternating GC-poor and GC-rich composition, as opposed to the genome of "cold-blooded" vertebrates (fishes and amphibians) that were supposed to lack GC-rich isochores. These findings were explained by the thermodynamic stability hypothesis, attributing genomic structure to body temperature. GC-rich isochores were purported to be a form of adaptation to environmental pressures, as an increase in genomic GC-content could protect DNA, RNA, and proteins from degradation by heat. Despite its attractive simplicity, the thermodynamic stability hypothesis has been repeatedly shown to be in error . Many authors showed the absence of a relationship between temperature and GC-content in vertebrates, while others showed the existence of GC-rich domains in "cold-blooded" vertebrates such as crocodiles, amphibians, and fish.

The isochore theory was the first to identify the nonuniformity of nucleotide composition within vertebrate genomes and predict that the genome of "warm-blooded" vertebrates such as mammals and birds are mosaic of isochores (Bernardi et al. 1985). The human genome, for example, was described as a mosaic of alternating low and high GC content isochores belonging to five compositional families, L1, L2, H1, H2, and H3, whose corresponding ranges of GC contents were said to be <38%, 38%-42%, 42%-47%, 47%-52%, and >52%, respectively.

The main predictions of the isochore theory are that:

Two opposite explanations that endeavored to explain the formations of isochores were vigorously debated as part of the neutralist-selectionist controversy. The first view was that isochores reflect variable mutation processes among genomic regions consistent with the neutral model. Alternatively, isochores were posited as a result of natural selection for certain compositional environment required by certain genes. Several hypotheses derive from the selectionist view, such as the thermodynamic stability hypothesis and the biased gene conversion hypothesis. Thus far, none of the theories provides a comprehensive explanation to the genome structure, and the topic is still under debate.

The isochore theory became one of the most useful theories in molecular evolution for many years. It was the first and most comprehensive attempt to explain the long-range compositional heterogeneity of vertebrate genomes within an evolutionary framework. Despite the interest in the early years in the isochore model, in recent years, the theory's methodology, terminology, and predictions have been challenged.

Because this theory was proposed in the 20th century before complete genomes were sequenced, it could not be fully tested for nearly 30 years. In the beginning of the 21st century, when the first genomes were made available it was clear that isochores do not exist in the human genome nor in other mammalian genomes. When failed to find isochores, many attacked the very existence of isochores. The most important predictor of isochores, GC3 was shown to have no predictable power to the GC content of nearby genomic regions, refuting findings from over 30 years of research, which were the basis for many isochore studies. Isochore-originators replied that the term was misinterpreted as isochores are not "homogeneous" but rather fairly homogeneous regions with a heterogeneous nature (especially) of GC-rich regions at the 5 kb scale, which only added to the already growing confusion. The reason for this ongoing frustration was the ambiguous definition of isochores as long and homogeneous, allowed some researchers to discover "isochores" and others to dismiss them, although both camps used the same data.