Data included observations on more than 200,000 first lactations of Norwegian cattle. Milk yield, fat and protein percentages, and observations on mastitis, ketosis, and presence of disease (binary coding of 0 or 1) were analyzed. Following Bayesian principles and applying the threshold concept, dispersion parameters for the binary traits (on the underlying scale) with continuous production traits were estimated. Heritabilities were .27, .34, and .43 for milk yield, fat percentage, and protein percentage, respectively. Heritabilities for the disease traits were .05 to .13 but may be inconsistent because of methodology problems with small sire by herd-year-season subclass size. Genetic correlations between milk yield and all three disease traits were above .5, indicating an undesirable relationship. Genetic correlations between ketosis and the content traits were -.38 to -.65; low component percentages were associated with high ketosis frequencies. Ignoring diseases in breeding programs may lead to undesirable correlated selection responses when selecting on milk yield.