Many chromosome-type, exchange-type chromosomal aberrations produced by radiation are intrachanges, i.e. involve only one chromosome. It is assumed such intrachanges are formed by illegitimate reunion of two double-strand breaks (DSBs) on the chromosome. The yield of intra-arm intrachanges (acentric rings or paracentric inversions) relative to that of interarm intrachanges (centric rings or pericentric inversions) is larger than would occur if production and illegitimate reunion of DSBs were spatially random. The excess of intra-arm intrachanges is presumably due to proximity effects for illegitimate reunions, i.e. enhancement of the intrachange probability when two DSBs are formed close to one another. Radiation track structure may also play a role. Using a polymer description for "large-scale" chromatin geometry (>2 Mb), and using two alternate (rapid or slow motion) models for the way that DSBs move after they are produced, theoretical estimates are given for size distributions of intrachanges at low or high linear energy transfer (LET). The ratio of intra-arm to interarm intrachanges is derived from the size distribution and compared with data from the literature on centric rings, inversions, interstitial deletions and excess acentric fragments. Proximity effects enhance yields of intra-arm relative to interarm intrachanges at least severalfold and perhaps as much as 10-fold compared to expectations based on spatial randomness. We argue that further measurements of intra-arm and interarm intrachanges would be informative about large-scale chromatin structure and chromosome motion. Because inversions are more frequent than estimates of randomness would indicate, and are transmissible to daughter cells, their size distribution could also help characterize past exposure to high-LET radiation.