SUMMARY Methods of utilising breeds and bleed crosses in animal production are discussed, taking account of both genetical and economic aspects. The theoretical principles for breed and breed cross comparison are analysed, but most emphasis is given to methods of improvement of existing crosses. A new synthetic breed is likely to have higher genetic variation, and reach a higher selection limit than the pure breeds from which it originates. However, it may take many years for the synthetic to surpass the best available purebred under continuous selection. Returns obtained in early years have more monetary benefit than those obtained later, for they can earn interest and incur a smaller risk element, so that a synthetic of use only in later years is unlikely to be cost-effective. Despite the flexibility in maintaining several alternative breeds, these need to be continually selected if they are to remain competitive, so better returns may be obtained by exerting more pressure on the best available present material. It is unlikely on theoretical grounds that cross testing schemes such as reciprocal recurrent selection have much to offer for breed cross improvement in large animals where growth and carcase traits are important.