Single genes can now be added routinely to the genome of mice by molecular manipulation as simple Mendelian dominants; this complements the normal process of reproduction to give 'transgenic' animals. Success in ruminants is limited to a few examples in sheep and although gene expression has yet to be documented, there is every reason to expect that it will be achieved. The application of this technology to livestock improvement depends on the identification of circumstances in which the phenotype is limited by the deficiency of a single protein. While there is little evidence to indicate that single dominant genes are in general likely to have favourable effects, it is argued that there are likely to be exceptions. These include particular combinations of promoter and structural gene sequences to alter feedback control, for example through a change in tissue specificity, and the alteration of definitive proteins such as those of milk. A mouse model has been established to study the molecular manipulation of sheep milk proteins. The sheep beta lactoglobulin gene has been incorporated and the sheep whey protein is secreted by the mammary gland of transgenic mice. For the future, means to delete or reduce the expression of existing genes are likely to be important, as are more effective means of incorporation such as retroviral based methods and the incorporation of multigene constructs. The resources required to test transgenic livestock will, however, be greater than those required to create them.