Chromosomal aberrations in somatic cells link DNA damage with radiation-induced cell killing and individual susceptibility to oncogenesis, and are also potential markers of cancer susceptibility. While there is general acceptance that the DNA double-strand break (DSB) is the principal initiating lesion the complexity of the relationship between the induced frequency and the rates of repair and misjoining of DSB, and the production of chromosome and chromatid aberrations has led to much controversy. The principal models of chromosome aberrations are: the classical 'breakage-and-reunion' or 'breakage-first' model of Sax [Genetics 25 (1940) 41-68], the 'mis-recombination' model of Chadwick and Leenhouts [Mutat Res 404 (1998) 113-117] and the 'transcription-based' model of Radford [Int J Radiat Biol 78 (2002) 1081-1093]. Chromatid aberrations have also been variously interpreted on the 'breakage-first model', Revell's 'exchange' model [Proc R Soc B 150 (1959) 563-589] and the 'signal' model [Int J Radiat Biol 73 (1998) 243-251]. Recent evidence argues strongly for different mechanisms for chromosome (formed in G1 or Go) and chromatid (formed in G2) aberrations, i.e. there is little or no correspondence in the relative frequencies between chromosome and chromatid aberrations. The balance of evidence indicates that chromosome aberrations may be formed by a breakage-first type mechanism. Elevated frequencies of chromosomal aberrations occur to various extents in cell lines mutated in genes involved in both non-homologous DSB end-joining and homologous recombinational rejoining of DSB. Chromatid breaks, seem to be formed by a more complex mechanism since there is a lack of correspondence between the rates of DSB rejoining and chromatid break 'disappearance' (assumed by some to represent DSB repair). Thus, a model based on the dissociation of DSB rejoining from chromatid break rejoining is required to explain these data. A substantial proportion (approximately 20%) of both spontaneous and induced chromatid breaks visibly involve inter-chromatid rearrangements (determined using harlequin staining of chromatids). It is postulated that the remaining proportion may also involve rearrangements, but within a single chromatid (i.e. intra-chromatid rearrangements). Disappearance of chromatid breaks with time is postulated to result from the completion of rearrangements, i.e. rather than simply from repair of DSB.