The objective of the present study was to examine if near infrared reflectance spectroscopy (NIRS) could be used to predict degradation parameters and effective degradation from scans of original forage samples. Degradability of dry matter (DM), crude protein (CP) and neutral detergent fibre (NDF) of 61 samples of perennial ryegrass (Lolium perenne L.) and orchardgrass (Dactylis glomerata L.) was tested by using the in situ technique. The grass samples were harvested at three different stages, early vegetative growth, early reproductive growth and late reproductive growth. Degradability was described in terms of immediately rumen soluble fraction (a fraction, for DM and CP only as NDF does not contain a soluble fraction), the degradable but not soluble faction (b fraction) and the rate of degradation of the b fraction (c value). Overall effective degradability of DM, CP and NDF was also calculated. Near infrared reflectance spectroscopy was examined for its ability to predict degradation parameters and to make a direct prediction of effective degradation from scans of the original samples of perennial ryegrass and orchardgrass. Prediction of effective degradation of the different feed fractions showed different accuracy. The coefficients of determination (R(2)) from regressions of predicted vs. measured effective degradation, using a cross-validation method, were 0.92 for DM, 0.78 for CP and 0.61 for NDF. The attempt to predict the degradation parameters (a, b and c) by NIRS was less successful as the coefficients of determination for the degradation parameters were low. Concentrations of CP and NDF in the original samples were predicted by using NIRS and the validated R(2) value was 0.98 for CP and 0.92 for NDF. It is concluded that using NIRS predictions from scans of original samples is a promising method to obtain values for the effective degradation of DM, CP and NDF in ruminant feeds, but that larger calibration sets are necessary for obtaining improved accuracy.