The existence of polyploid mammalian spermatozoa has been inferred from studies of Feulgen-DNA absorption. Rabbit spermatozoa fell into two discrete groups with mean absorptions close to a 1:2 ratio (inferred to be haploids and diploids respectively) ; simple visual appraisal of the size of the head or nucleus gave an identical classification. The incidences of ploidy classes were 98\m=.\94%haploid, 1\m=.\06%diploid, 0\m=.\00%higher than diploid (N= 3010; from DNA measurements and visual appraisal of the size in a rabbit chosen to have a high incidence of diploids) and, correspondingly, 99\m=.\691%,0\m=.\308%,0\m=.\001% (N=138001; from sixty-nine unselected rabbits, scored by visual appraisal of the size of the sperm head). In man also, virtually discrete groups with absorptions close to a 1:2 ratio existed and were inferred to be haploids and diploids respectively. A few human spermatozoa were found with absorptions corresponding to a ploidy of three and/or four. Visual appraisal of the size of the human sperm nucleus as Small, Medium or Large was only a partial guide to ploidy. All Small human spermatozoa measured for DNA absorption were found to be haploid. About two-thirds of Medium human spermatozoa were found, however, to be haploid, and some Large spermatozoa were haploid or diploid. The incidences of ploidy classes in the human were 99\m=.\37%haploid, 0\m=.\56%diploid, 0\m=.\07%higher than diploid (N=5554; with consistency between duplicate slides and between two subjects; from DNA measurements and visual appraisal of nuclear size). The estimated incidence of diploid human spermatozoa is consistent with the known incidence of triploid fetuses. In a mouse with a putatively high incidence of diploids, all 1000 DNA measurements were nevertheless within the haploid range, with one diploid encountered outside the main sampling.