Autotetraploid gene segregation was studied in Zea mays L. using a marking system of two very closely linked genes (A 1 and Sh 2) in the repulsion phase. This system makes it possible to identify many euploid and aneuploid genotypes and enables the estimation of some parameters of autotetraploid gene segregation such as double reduction, numerical nondisjunction, and the relative transmission frequencies of monosomic, disomic, and trisomic gametes. It was found that these three types of gametes did not function at the same rates on the male and female sides. Differences in observed segregation ratios between reciprocal testcrosses were explained by this phenomenon. Estimates of the frequency of double reduction were made for loci used after eliminating the effect of numerical non-disjuction on the segregation ratios. The value of double reduction appears to be the same in the male and female tetrasomic tetraploid. Tetraploids which were disomic for chromosome 3 were not isolated although they might be expected to be common in the progeny of self-fertilized or sib-crossed trisomic tetraploids. Their absence may be explained in part by the low rate of transmission of monosomic gametes from the male parent. Autotetraploid populations which are unstable for chromosome number probably achieve an equilibrium between forces which produce aneuploidy and forces which remove aneuploids from the population.