SUMMARYMoulds belonging to Penicillium section roqueforti are common contaminants of feedstuffs and produce several mycotoxins that can cause health hazards when ingested by farm animals. Among these, PR toxin (PR), mycophenolic acid (MY) and roquefortine C (RC) have been frequently detected in forages, particularly silages. The aims of the current trials were to study the effects of the presence of pure mycotoxins on in vitro rumen fermentation parameters and to assess their stability in the rumen environment. Two successive in vitro gas production experiments were carried out: a central composite design with four replications of central point (CCD) and a completely randomized design with a fully factorial arrangement of treatments (FFD). In CCD, the effects of PR, MY and RC concentrations in diluted rumen fluid (i.e. 0·01, 0·30, 1·01, 1·71 and 2·00 μg of each mycotoxin/ml) were tested. Gas volume produced after 48 h of incubation (Vf) decreased linearly as concentrations of RC and MY in diluted rumen fluid increased, with marginal effects similar for two mycotoxins, being respectively −14·6 and −13·4 ml/g organic matter (OM) for each 1·0 μg/ml of increment in mycotoxin concentration. Similarly, total volatile fatty acid (VFA) production decreased quadratically as concentrations of RC and MY increased, with marginal effects about two times higher for MY than RC, being −4·22 and −2·62 mmol/l for each 1·0 μg/ml of increment in mycotoxin concentration. With respect to maximum Vf (i.e. 410·6 ml/g OM) and VFA (98·06 mmol/l) values estimated by the model, decreases of 13·6 and 15·2% were obtained when incubating the highest RC and MY concentrations, respectively. The PR did not interfere with rumen fermentation pattern and it was not recovered after 48 h of incubation, whereas the stabilities of MY and RC in rumen fluid were similar and on average equal to about 50%. On the basis of CCD results, a second experiment (FFD) was carried out in which only effects of MY and RC concentrations (i.e. 0, 0·67, 1·33 and 2·00 μg of each mycotoxin/ml of diluted rumen fluid) were tested. Data from FFD showed Vf decreased linearly when concentrations of MY and RC increased, with marginal effect two-folds higher for MY than for RC (−11·1 ml/g OM and −6·7 ml/g OM, respectively). Similar marginal effects of MY and RC in decreasing VFA production were recorded: −2·38 and −2·86 mmol/l for each 1·0 μg/ml of increment in mycotoxin concentration, respectively. At the highest RC and MY tested concentrations, Vf and VFA decreased by 8·7 and 10·7%, respectively, over maximum estimated values. In FFD, the average amounts of MY and RC recovered in rumen fluid after 48 h of incubation were 79·0 and 40·6%, respectively. In conclusion, the MY and RC from standards interfered with rumen microorganisms at relatively low levels and were partially stable in the rumen environment after 48 h of incubation. These findings suggested that MY and RC could interfere with digestive processes and might represent a potential risk for ruminants fed diets containing feeds contaminated by mycotoxins produced by P. roqueforti.