The objective of this study was to estimate the dominance variance for postweaning gain in Limousin cattle. Data included 215,326 records of postweaning gain from 205 to 365 d, provided by the North American Limousin Foundation. Parental dominance subclasses were formed and related using the method of Hoeschele and VanRaden. Variance components were estimated using Method R based on six samples of 50%. Fixed effects in the model included contemporary group and covariates for inbreeding and breed composition (percentage Limousin). Heterozygosity was negatively correlated with breed composition (< -.99) and was therefore not included in the model. Two types of contemporary groups used as original groups from the National Cattle Evaluation were partially based on breed composition. Original contemporary groups that were too homogeneous for breed composition were replaced by herd-year-sex classes. Two models were used with the two data sets. Model 1 contained the fixed effects described above and an additive genetic effect. Model 2 included a dominance effect in addition to the effects contained in Model 1. In total, four combinations of contemporary group x model were used. Dominance variance was computed as being four times the estimated parental subclass variance. Estimates for inbreeding depression and breed composition (percentage Limousin) were all small and not greatly affected by inclusion of dominance effects or changes in contemporary groups. Estimates of the additive variance (expressed as percentage of the phenotypic variance) were only slightly affected, with values between 20 and 21%. Dominance estimates were highly affected when passing from original (10%) and to alternative contemporary groups (18%). Such large values may indicate that dominance is important for postweaning gain. Results showed the advantage of an individual dominance approach based on sire-dam combinations; therefore, expected gains through the use of specific combination ability as a part of the mating selection criteria for growth might be high.