Adding sugar to the diet has been reported to improve production in dairy cows. In each of 2 trials, 48 lactating Holsteins (8 with ruminal cannulas) were fed covariate diets for 2 wk, blocked by days in milk into 12 groups of 4, and then randomly assigned to diets based on alfalfa silage containing 4 levels of dried molasses (trial 1) or liquid molasses (trial 2). In both studies, production data were collected for 8 wk, ruminal samples were taken in wk 4 and 8, and statistical models were used that included covariate means and block. In trial 1, experimental diets contained 18% CP and 0, 4, 8, or 12% dried molasses with 2.6, 4.2, 5.6, or 7.2% total sugar. With increasing sugar, there was a linear increase in dry matter intake (DMI), and digestibility of dry matter (DM) and organic matter (OM), but no effect on yield of milk or protein. This resulted in linear decreases in fat-corrected milk (FCM)/DMI and milk N/N-intake. There was a linear decrease in urinary N with increasing sugar, and quadratic effects on milk fat content, yield of fat and FCM, and ruminal ammonia. Mean optimum from these quadratic responses was 4.8% total sugar in these diets. In trial 2, experimental diets contained 15.6% crude protein (CP) and 0, 3, 6, or 9% liquid molasses with 2.6, 4.9, 7.4, or 10.0% total sugar, respectively. Again, there were linear declines in FCM/DMI and milk N/N-intake with increasing sugar, but quadratic responses for DMI, yield of milk, protein, and SNF, digestibility of neutral detergent fiber and acid detergent fiber, milk urea, urinary excretion of purine derivatives, and ruminal ammonia. Mean optimum from all quadratic responses in this trial was 6.3% total sugar. An estimate of an overall optimum, based on yield of fat and FCM (trial 1) and yield of milk, protein, and SNF (trial 2), was 5.0% total sugar, equivalent to adding 2.4% sugar to the basal diets. Feeding more than 6% total sugar appeared to depress production.