Twenty-four low dry matter (DM) silages differing in fermentation quality were harvested at the same time from a crop that consisted mainly of timothy (Phleum pratense), and meadow fescue (Festuca pratensis). The silage samples were analysed by gas chromatography (GC) - mass spectrometry and gas chromatography - flame ionisation detection in order to determine and quantify volatiles present in silage. The voluntary intake of the 24 silages had been measured in a previous feeding trial with growing steers of Norwegian Red. Thirteen esters, five aldehydes, three alcohols, and one sulphide were identified and quantified. A total of 51 variables describing the chemical composition of the silages were included in a partial least-squares regression, and the relationship of silage fermentation quality to voluntary intake was elucidated. The importance of variables describing silage fermentation quality in relation to intake was judged from a best combination procedure, jack-knifing, and empirical correlations of the variables to intake. The GC-analysed compounds were mainly present in poorly fermented silages. However, compared with other explanatory chemical variables none of these compounds was of importance for the voluntary intake as evaluated by partial least-squares regression. A validated variance of 71% in silage DM intake was explained with the selected variables: total acids (TA), total volatile fatty acids (TVFA), lactic acid/total acid ratio and propionic acid. In this study extent (by the variable TA) and type of silage fermentation (by TVFA) influenced intake. Further, it is suggested that by restricting the fermentation in low DM grass silages the potential intake of silage DM is maximised.