Implementation of genomic selection requires a large reference population to accurately estimate SNP effects. Some Canadian dairy breeds do not have large enough reference populations to estimate SNP effects accurately. If the gametic phase of adjacent marker is highly correlated across multiple breeds, it would be possible to pool breeds into one common reference population. This study investigated the amount of linkage disequilibrium (LD) in 5 major dairy breeds using the 50k SNP panel (Brown Swiss(BS), Ayrshire(AY), Guernsey (GU), Holstein(Ho), Jersey (JE)) and in 3 of the same breeds using the 777k SNP panel (AY, GU, HO). Both the extent of LD at varying distances and the correlation of pair-wise SNP gametic phase were investigated. SNPs were filtered for minor allele frequency (<5%). The level of LD was measured using the squared correlation between alleles at 2 loci (r2), and consistency of SNP phase was assessed using the signed root of these values. LD values greater than 0.2 were found in all breeds at the average pair-wise distance between adjacent SNP using the 50k panel. Correlations of r values, however, did not reach high levels at this distance (r=0.48 to 0.59). High correlation values of SNP phase between breeds were found when the average pair-wise distances between adjacent SNP using the 777k SNP panel were examined. In this case, the phase correlation varied from 0.93 to 0.96 between the 3 breeds. These findings suggest that a multi-breed reference population might be advantageous when the 777k SNP panel is used.