Twenty multiparous Jersey cows, paired according to previous lactation yield and expected date of calving, were given two diets in a cross-over experiment in early lactation. The diets had forage:concentrate ratios maintained at 3:7, the forage being wilted grass silage and the concentrate either a proprietary dairy cake or the same cake containing saliva salts (SS) at 40 g/kg dry matter. The SS diet proved acceptable to the animals and had no effect on gestation length or calf survival. Absolute milk yields were not affected, but the SS diet caused a significant increase in the proportions of milk fat, total solids (both P < 0·001) and protein (P < 0·05) and in the yield of fat-corrected milk (P < 0·001). Analysis of rumen liquor samples showed that giving the SS diet resulted in a significant increase in the molar proportion of acetate and a decrease in the proportions of propionate and to a lesser extent butyrate. Milk fat content showed significant correlations with ruminal acetate (r = 0·67), propionate (r = −0·59) and the acetate: propionate ratio (r = 0·64).