Hydrolysable tannins (HT) show potential as silage additive for autumn herbage silages, high in (rumen degradable) protein, as they may reduce proteolysis. Additionally, they have abilities to form pH‐reversible tannin–protein complexes, non‐degradable in the rumen but degradable in the abomasum and intestines of ruminants. Therefore they can improve milk N efficiency and shift N excretions from urine to faeces, possibly mitigating the environmental impact of ruminants. In this study, two small bunker silos were filled with autumn grass. One was treated with 20 g/kg DM HT extract (TAN) (TannoSan‐L), the other with 8 mg/kg DM inoculant containing lactic acid bacteria (INO) (Bonsilage Fit G). Secondly, micro‐silos (2.75 L) were filled with four treatments; (1) grass without additive (CON) (n = 5); (2) TAN (n = 5); (3) INO (n = 5); and (4) TAN + INO (n = 5). The bunker silos were used in a cross‐over feeding experiment with periods of 4 weeks involving 22 lactating Holstein cows (average ± SD: 183 ± 36.3 days in milk, 665 ± 71.0 kg body weight, and 33.8 ± 3.91 kg/day milk yield). The HT dose was insufficient to reduce proteolysis or alter chemical composition and nutritional value in the micro‐ and bunker silages. Including grass silage added with TAN (3.2 g HT/kg DM) in the diet, did not affect feed intake nor fat and protein corrected milk yield in comparison to feeding the grass silage added with INO in a similar diet. The TAN‐fed cows had an increased faecal N excretion and decreased apparent total‐tract N and organic matter digestibility, but no improvement in the cows’ N utilization could be confirmed in milk and blood urea levels. Overall, feeding an autumn grass silage treated with 20 g/kg chestnut HT extract did not affect the performance of dairy cows in comparison to feeding an autumn grass silage treated with a lactic acid bacteria inoculant.