Objective genetic improvement in specialised meat breeds of sheep in Britain is based largely on the performance recording and genetic evaluation service provided by Signet. This includes the use of ultrasonic scanning and a selection index to identify animals with high genetic merit for lean growth. Substantial responses to selection (about 2% per annum in the unscaled index) have been achieved in an experimental flock in which selection was based on this index. There is also good experimental evidence that sires selected on this index produce purebred and crossbred progeny with improved carcass lean:fat ratio, at a range of degrees of maturity, and in a range of feeding systems. Over 20 sire referencing schemes have been established in Britain during the last 12 years, mainly in specialised meat breeds. These schemes account for about half of the performance-recorded flocks in Britain. The larger schemes involve around 6,000-7,000 ewes in 60-80 flocks. These co-operative breeding schemes create genetic links between flocks and years through the shared use of elite rams (reference sires), often via artificial insemination. Multi-trait animal model BLUP is then used to provide across-flock genetic evaluations. These in turn allow increased selection intensities and increased response to selection. In the specialised meat breeds the lean growth index mentioned above is a primary selection criterion. High responses to selection for lean growth (about 1.75% per annum in the larger schemes) and other measures of performance are being achieved. These schemes also provide an ideal structure for the cost-effective adoption of new technologies, such as advanced methods for in vivo estimation of carcass composition, and molecular genetic markers. The formation and growth of these schemes has been stimulated by (i) the national availability of relevant performance-recording services/selection criteria, (ii) the improved success of laparoscopic artificial insemination with frozen semen, (iii) the wider availability of genetic evaluation software and powerful computers, and (iv) research, and interaction with breeders, on the optimal design and operation of schemes, including a measure of connectedness among flocks.