Genetic parameters were estimated for purebred and crossbred lambs of a local breed using a single trait animal model. The traits recorded were: birth weight (BWT), weight at 30 days (WT30), weaning weight (WWT), average daily gain from 1 to 30 days (ADG1) and average daily gain from 30 to 90 days (ADG2). Five different animal models were fitted for each trait; all including additive direct genetic variance and various combinations of additive maternal and environmental maternal effects. The most appropriate model was chosen based on likelihood ratio tests. Additive maternal and permanent environmental effects were important (P < 0·05) for birth weight in purebred and crossbred lambs, when compared with a model containing only additive direct effects. Inclusion of maternal permanent environmental effects provided a better fit (P < 0·05) for weaning weight in purebred lambs than a model containing only additive direct effects. Estimates of heritability from the model containing additive direct, maternal effects and maternal permanent environmental effects, but not additive direct-maternal correlation, for combined purebred and crossbred lambs were 0·32 for BWT, 0·19 for WT30, 0·24 for WWT, 0·26 for ADG1 and 0·12 for ADG2. Estimates of additive maternal and maternal permanent environmental variances, respectively, as a proportion of phenotypic variance were 0·06 and 0·07 for BWT, 0·05 and 0·02 for WT30, 0·02 and 0·03 for WWT, 0·03 and 0·05 for ADG1 and 0·00 and 0·03 for ADG2. Estimates of direct-maternal correlation in subsequent analyses were significant and ranged from −0·16 to −0·95 for live weights and from −0·73 to −1·0 for daily gains. However, the very large negative correlations probably resulted from undefined non-genetic covariances as well as possible antagonistic genetic effects. These results indicate that it would be possible to improve growth traits in a local sheep breed through genetic selection.