Intrabreed additive genetic, environmental, and phenotypic variances and covariances for Romosinuano (R) and Zebu (Z), as well as interbreed nonadditive genetic variances and covariances (sire x breed-group-of-dam interactions), for birth weight (BWT), weaning weight (WW), and postweaning gain (GW) were computed using a Romosinuano-Zebu multibreed data set from the Turipaná Experiment Station in Colombia. Covariances were estimated with a sire-maternal grandsire model, using a multibreed REML procedure. The computing algorithm was a generalized expectation-maximization (GEM) algorithm. This algorithm yields no asymptotic standard errors as part of its computations. Because of the small size of the data set (2,546 calves), these REML covariance estimates should be viewed with caution. Estimates of intrabreed heritabilities were similar to the ratios of interbreed nonadditive to phenotypic variances (interactibilities) for direct and maternal effects of the growth traits evaluated in this study. Intrabreed heritability estimates for BWT, WW, and GW direct genetic effects were .16, .09, and .14, for R and .24, .10, and .14 for Z. Corresponding heritability values for maternal effects were .18, .09, and .23 for R and .14, .13, and .07 for Z. Interactibility estimates were .21, .05, and .12, for direct and .26, .04, and .11, for maternal BWT, WW, and GW. Negative correlations between additive maternal weaning weight and direct and maternal postweaning gain, as well as between environmental weaning weight and postweaning gain, suggested that there was compensatory postweaning gain in this herd. Estimates of genetic variation and predictions of expected progeny differences showed that Romosinuano animals competed well against Zebu and RZ crossbred animals under the tropical environmental conditions at Turipaná.