AbstractA randomized block design experiment involving 64 lactating dairy cattle was carried out to evaluate a bacterial inoculant and two rates of formic acid as silage additives. Herbage from the primary growth of a predominantly perennial ryegrass sward was ensiled unwilted using two precision-chop harvesters. Alternate loads of herbage were either untreated (UT) or treated with formic acid applied at the rate of 2·7 l/t grass (LF) and 5·7 l/t grass (HF) or an inoculant enzyme preparation applied at the rate of 2·2 l/t grass (I). The mean dry matter (DM) and water-soluble carbohydrate (WSC) concentrations and buffering capacity of the UT herbage were 148 g/kg, 151 g/kg DM and 638 mEq/kg DM. Inoculant treatment had no effect on the rate of silage fermentation post ensiling relative to the UT silage whereas HF treatment produced a restricted fermented silage. For silages UT, LF, I and HF, pH values were 4·53, 3·89, 4·32 and 4·05 (s.e.d. = 0·061); ammonia nitrogen (N) concentrations were 179, 89, 150 and 73 g/kg total N (s.e.d. = 10·4); butyrate concentrations were 6·0,1·9, 5·1 and 2·0 g/kg DM (s.e.d. = 1·24); and WSC concentrations were 14·4, 35·2, 18·0 and 100·7 g/kg DM (s.e.d. = 1·24), respectively. The silages were offeredad libitum. The UT silage was supplemented with either 2 (UT2), 4 (UT4) or 6 (UT6) kg concentrates per head daily while the LF, I and HF silages were supplemented with 4 kg (LF4,14, HF4) of concentrates. For treatments UT4, LF4, 14 and HF4, silage DM intakes were 9·69, 10·89, 9·93 and 10·89 (average s.e.d. = 0·465) kg/day; fat plus protein yields 1·39,1·44,1·32 and 1·49 (average s.e.d. = 0·057) kg/day; protein concentrations 29·2, 31·1, 29·9 and 30·2 (average s.e.d. = 1·01) g/kg. Inoculant treatment increased N apparent digestibility (P < 0·05) of the total diets relative to UT. The apparent digestibilities of neutral-detergent fibre (P < 0·05) and hemicellulose (P < 0·01) were decreased due to HF treatment. For the production of a constant milk fat plus protein yield it was estimated that an extra 0·63 kg per head per day and 1·04 and 1·77 kg per head per day less concentrate would be required for the I, LF and HF treatments, respectively. It is concluded that relative to the UT treatment, I treatment had no beneficial effects on silage fermentation, total diet apparent digestibility (other than N apparent digestibility) or animal performance. The LF treatment improve lage fermentation and intake, and tended to increase the yield of fat plus protein (P > 0·05) whereas HF treatment restricted silage preservation and increased silage intake and the yields of protein and fat plus protein.