Bacterial mastitis (inflammation of the mammary gland) is the most frequent and costly disease of dairy cattle. As disease-recording systems compiling data from large numbers of farms are not widely implemented, selection for mastitis resistance is often based on genetically correlated traits, including somatic cell count (SCC), udder depth and fore udder attachment. However, in Canada, a national health collection system for dairy cattle was started in 2007, with producers recording the incidence of 8 diseases, including mastitis. Participation is voluntary, but already includes >40% of all Canadian herds enrolled in milk recording. In several studies using these Canadian data, various traits with potential to predict mastitis resistance were investigated. The relation between mastitis and SCC (and its variations during lactation) were also studied, with the most important being mean somatic cell score (SCS) in early lactation, standard deviation of SCS and excessive test-day SCC pattern. Genetic correlations between mastitis and other traits were also estimated. Selection for mastitis resistance would also improve resistance against other diseases, as well as enhance both fertility and longevity. However, milk yield was negatively correlated with clinical mastitis, emphasizing the importance of including health traits in genetic selection. Based on these studies, routine genomic evaluation for mastitis resistance was initiated in Canada (August 2014). The new evaluation incorporated mastitis, patterns of SCC, udder depth, fore udder attachment and body condition score. By including predictor traits in multiple-trait evaluations, EBV reliability increased 22 points. Furthermore, with genomics, reliability increased by 16 and 8 points for young and proven bulls, respectively. Further research, including the bacteriology of mastitis and development of a new female reference population to improve reliability of genomic evaluation, are still under study and should result in further improvements.