The most commonly accepted view about the origin of aneuploidy is that it is due to errors in meiotic division. However, its rare occurrence makes it difficult to explain recurrent births of trisomic children to some parents. This problem causes more serious concern when one accepts that an abnormal (n + 1 or n - 1) sperm would enter fertilization by overriding thousands, or even millions, of normal haploid sperms. Also, the failure of aneuploidy to be induced in the offspring of mammals treated with mutagens raises questions about the effectiveness of the accepted mode of origin of errors. Current concepts also do not explain why one observes more errors of meiotic I, than of meiotic II, origin. It is known that most chromosomes separating at meta-anaphase junction in mitosis follow a nonrandom, genetically controlled sequence of separation. The present proposal makes use of out-of-phase separation of a rare chromosome, like premature separation in mitosis of the X in elderly humans or of an 18 in parents of trisomy 18 children. The suggestion is made that such out-of-phase separation results in aneuploid cell lines by total failure of the centromere to separate or by it separating too early, before the spindle is formed. The prematurely separating centromeres, it appears, do not attach to spindle fibers and hence cause nondisjunction. Such nondisjunction in embryonic stages will produce apparently normal individuals with mosaicism in somatic and/or gametic tissue. An individual carrying mosaicism in gonadal tissue will produce a large number of abnormal gametes, one of which may have a reasonable chance of entering fertilization. This mode of origin of aneuploidy takes care of all questions raised above and finds support in the data available in the literature. Several of the suggestions made in the hypothesis are easily testable.