This project exploits two novel technologies to address genetic improvement of complex, health traits in pedigree dogs: the development of dense canine SNP panels and genomic selection, a statistical approach that uses such genotype information to provide genomic breeding values for use in a selection scheme. This approach is currently being applied in dairy cattle but has not previously been used for pedigree dogs. We plan to apply this approach to predisposition to hip dysplasia (as assessed by hip score) in the Labrador Retriever, as a test case. Our primary goals are to calculate genomic breeding values and to identify genomic regions associated with hip score. To achieve this, we will genotype 1500 animals from the extremes of the phenotypic distribution in Labradors, using the Illumina 200K CanineHD array. Based on theoretical approximations and heritability estimates for this disease, we predict that this approach should result in faster selection for improved health than a traditional phenotype- and pedigree-based approach. The genomic regions identified in the study will be further analysed by comparison with results from previous QTL studies of hip dysplasia in dogs, with the aim of identifying candidate genes. We will also analyse genetic interactions between hip score and other traits, including elbow score (associated with the related disease, elbow dysplasia). Our secondary objective is to develop an implementation plan for genomic selection against hip dysplasia. This will involve identification of lower-density SNP panels that characterise maximal genetic variation for given numbers of SNPs. We will also perform a simulation study to evaluate the efficacy of alternative breeding and genotyping schemes. Several features will be evaluated for their effects on genetic progress: (1) size and breadth of the genotyping panel, (2) timing of genotyping, and (3) proportion of dogs genotyped in the breed.