Four experiments were carried out in Lelystad, The Netherlands in 1994, in which perennial ryegrass wilted to 421–568 g dry matter (DM) kg−1 was ensiled with and without an inoculant containing Lactobacillus plantarum and Enterococcus faecium strains in 1-litre capacity laboratory silos. Treated silages showed a markedly increased rate of pH decline. The final pH of treated silages was reached 20–30 days after ensiling, whereas acidification of control silages continued during the full 180-day ensilage period. After 180 days ensilage, treated silages showed significantly (P<0·001) lower pH, DM loss and ammonia-N concentrations and significantly (P<0·001) higher lactic acid concentrations than control silages in all experiments. Concentrations of ethanol and acetic acid were significantly (at least P<0·05) lower for the treated silages, except for acetic acid in one experiment and ethanol in another. Butyric acid was not found in any of the silages. One of the control silages (ensiled at 517 g DM kg−1) contained ethanol as the major fermentation product indicating that alcoholic fermentation had taken place, probably by yeasts. Treated and untreated grasses used in the first experiment (432 and 442 g DM kg−1, respectively) were also used to produce silages in two 40-t capacity clamp silos. Similar to the laboratory silos, the treated silage had a higher lactic acid concentration and lower pH and ethanol, acetic acid and ammonia-N concentrations than the untreated silage.