An in situ study was conducted on four rumen-cannulated wethers to determine (using 15N infusion techniques) the microbial contamination (mg bacterial DM or crude protein (CP)/100mg DM or CP) and the associated error on the effective degradability of fourteen feeds: barley and maize grains, soyabean and sunflower meals, full-fat soyabean, maize gluten feed, soyabean hulls, brewers dried grains, sugarbeet pulp, wheat bran, lucerne and vetch-oat hays, and barley and lentil straws. The DM or CP contamination in residues (M) fitted to single exponential or sigmoid curves. A general model (M=m (1−e−ft)j) was proposed to match this fit. Asymptotic values (m) varied from 2·84% to 13·3% and from 2·85% to 80·9% for DM and CP, respectively. Uncorrected results underestimated the effective degradability of both DM (P<0·05) and CP (P<0·01). For CP, this underestimation varied from 0·59% to 13·1%, with a higher but unascertainable error for barley straw. Excluding maize grain, the microbial contamination of both DM and CP, and the associated underestimation of the effective degradability of CP, were positively related to the cellulose content of the feed. The error in the effective degradability of CP was also negatively related to the CP content and its apparent effective degradability (R2 0·867). This equation allows easier and more accurate estimates of effective degradability, needed to improve protein-rationing systems.