A 5-year-old male with mild mental retardation showed a chromosomal rearrangement involving duplication of part of 2q (2q33.3 leads to 2wter) in 70% of metaphases from peripheral blood; the remaining 30% of cells had a rearrangement of chromosome 2 in the form of a ring, viz., r(2)(2p25.2 leads to 2q33.2). Both configurations appeared to be missing a tiny portion of 2p (p25.3 leads to pter). All metaphases examined from cultured skin fibroblasts from the child had the abnormal rod configuration of chromosome 2; none had the ring. The pattern of the 2q duplication/2p deletion in the rod is that expected if there were an inversion in a No. 2 in one of the parents. Q-, G-, and R-banding studies, however, revealed both parents to be chromosomally normal. Furthermore, the finding of an inversion would not explain the origin of the ring. The most probable explanation is that neither parent is mosaic for an inversion, but that the rod and ring configurations arose simultaneously from a de novo, prezygotic or early zygotic exchange in a No. 2, either between complementary DNA strands in G1 or by intrachromosomal exchange in S or G2. Differential selection against cells with the ring chromosome in blood and skin probably occurred during subsequent embryological development. Cytoplasmic malate dehydrogenase (MDH1) was excluded from the terminal band of 2p (i.e., 2p25.3) by deletion mapping.