Banding studies in 25 Robertsonian translocations showed that all could be interpreted as stable dicentrics. The mechanism for their stability is likely to be the proximity of their centromeres but centromeric suppression could also have a role. In many of these dicentric translocations, discontinuous centromeric suppression, as indicated by chromatid separation at one of the centromeric regions, was observed in C-banded preparations. A further observation of undefined relation to the first was that the ratio of the two constitutive centromeric heterochromatin (CCH) regions from the component chromosomes of the translocations was variable in the same translocation type, e.g. t(13;14). It is proposed that this ratio may influence the segregation ratio. Abnormal spermatogenesis is suggested as the likely mechanism for the difference in the proportion of aneuploid offspring in the progeny of maternal and paternal heterozygotes. Neither of the t dic(21;21)s could be interpreted as isochromosomes. It is proposed that Robertsonian fusion translocations be defined as stable, dicentric, whole-arm translocations, with both centromeres in a median position and resulting in the loss of a small acentric fragment during this formation. It is suggested that they occur at high frequency between telocentric or, as in man, certain acrocentric chromosomes because of some intrinsic property of those chromosomes not possessed by metacentric chromosomes and mediated by interphase association of centromeres.