This experiment was conducted to evaluate the effects of improved protein quality of grass silage by additive use on dairy cow intake and performance. Grass from the first regrowth was wilted to 33% DM before being precision chopped and treated with the inoculant Kofasil Duo (Lactobacillus plantarum/Lactobacillus buchneri, 200,000 cfu/g) or with the chemical additive Kofasil Ultra K (sodium nitrite, hexamine, potassium sorbate, sodium benzoate and sodium propionate, 2 L/T, Addcon Europe GmbH), which were compared with a control without additive. The forage was ensiled in hard-pressed round bales before being fed to 48 dairy cows (150 DIM) in a continuous trial for 9 wk using 8 cows per treatment. Cows were blocked according to breed, parity, DIM and milk yield. The silage contained 15% CP, 47% NDF, 3.3% WSC, 8.2% lactic acid, 2.1% acetic acid and 0.25% NH3-N of DM with minor differences between treatments. Rumen undegradable protein (RUP) of the silage at 5% passage rate was 22, 23 and 24% of CP for control (C), inoculant (I) and salt-based (S) additive, respectively. Diets were isonitrogenous (15.1% of DM) and isoenergetic (11.4 MJ/kg DM) varying in RUP (4.7% (high) and 2.5% (low) of DM). Concentrate UDP was 2.9 and 8.6% of DM. Dietary forage proportion of the TMR was 58% of DM. Data were analyzed using PROC MIXED of SAS with RUP, silage and week as fixed factors and block as a random factor. High RUP diet had higher milk yield than low RUP diet (29.4 vs. 27.9 kg; P < 0.05). The DM intake was not affected by RUP and silage treatment. Yields of milk and ECM were higher for the I and S diets than for the C diet at low RUP (28.9 vs. 26.0 kg milk, P < 0.01; 30.6 vs. 27.1 kg ECM, P < 0.001) whereas there was no effect of additive treatment in the high RUP diet. Milk fat and protein did not differ between silage treatments. Feed efficiency was higher for the I and S diets than for the C diet at the low RUP (1.6 vs. 1.3 kg of ECM/kg DM intake, P < 0.001) but not at the high RUP. The increased milk yield and feed efficiency when fed a diet with low RUP can partly be explained by increased RUP of the I and S silages.