Foundation cows were selected using prior records from one of two sources, private herds or other projects at the Research Center. Comparing twinning rates before and after selection, the repeatability was lower for those from the first (.08) than for those from the second (.16) source with a combined value of .12. Realized heritability of single-parity twinning rate estimated from selection of parents and response in daughters of foundation females was .06. Paternal half-sib estimate of heritability of twinning rate was .02 +/- .07. Estimates of repeatability computed from calving records of females born in the project indicate that permanent environmental effects on twinning rate in cattle are small. Mean calving rate of females born in the project was 1.11 in the data set that included all data and 1.09 in the data set that excluded females from highly selected parents. Twinning rate was greater (P less than .05) in fall than in spring calving (1.13 vs 1.06). Data on twinning rate of a sample of the Swedish Friesian breed were summarized. Mean twinning rate of the Swedish Friesian breed is 2.57%. Age adjusted mean twinning rate of daughters of 32 half-sib sons of one particular Swedish Frieisan sire averaged 5.4% and ranged from .9% to 13.6%. There was no indication of a bimodal distribution, which would be expected if a single gene with a major effect on twinning rate were segregating. The estimated genetic standard deviation (sigma-xn) for mean twinning rates of the 32 sire progeny groups was 1.8%. Observed range among son progenies was .127 or 7.2 sigma-xn, in reasonable agreement with the hypothesis that twinning rate in this population is inherited as a quantitative trait.