The objective of this study was to investigate relationships between different cell wall (CW) fractions and in vivo digestibility of grass and red clover silages. The material consisted of 28 silages (6 primary growth grass, 16 regrowth grass and 6 red clover). Different fibre fractions predicted organic matter (OM) and neutral detergent fibre (aNDFom) digestibility reasonably well among the three forage types, but uniformly the digestibility predictions were not accurate enough for estimating the feeding value. Isolation of total CW did not improve digestibility predictions compared with aNDFom. The fraction insoluble in CW isolation, but soluble in aNDFom extraction was a uniform feed fraction with almost complete true digestibility. Isolation of CW fraction from faecal samples was not able to solubilize endogenous and metabolic components. True crude protein (CP) digestibility was higher than 0.90 when estimated by the Lucas test and by assuming that only N bound to neutral detergent fibre in faeces was of feed origin. Acid detergent lignin was more closely related to digestibility than lignin isolated from neutraldtergent fibre or CW (but all were inferior compared to indigestible NDF (iNDF) determined by 12-d ruminal in situ incubation. Lignin to iNDF ratio varied both within and between forage types. For evaluating nutritive value of forages actual measurements of iNDF (or undigested NDF) is recommended over detailed analysis of fibre and lignin fractions. Despite some CW fractions are solubilized in ND extractions, analysis of total CW is not a useful tool in ruminant nutrition research.