The production of extreme or 'transgressive' phenotypes in segregating hybrid populations has been speculated to contribute to niche divergence of hybrid lineages. Here, we assess the frequency of transgressive segregation in hybrid populations, describe its genetic basis and discuss the factors that best predict its occurrence. From a survey of 171 studies that report phenotypic variation in segregating hybrid populations, we show that transgression is the rule rather than the exception. In fact, 155 of the 171 studies (91%) report at least one transgressive trait, and 44% of 1229 traits examined were transgressive. Transgression occurred most frequently in intraspecific crosses involving inbred, domesticated plant populations, and least frequently in interspecific crosses between outbred, wild animal species. Quantitative genetic studies of plant hybrids consistently point to the action of complementary genes as the primary cause of transgression, although overdominance and epistasis also contribute. Complementary genes appear to be common for most traits, with the possible exception of those with a history of disruptive selection. These results lend credence to the view that hybridization may provide the raw material for rapid adaptation and provide a simple explanation for niche divergence and phenotypic novelty often associated with hybrid lineages.