In the Dutch protein evaluation system (Tamminga et al., 1994) microbial protein synthesized in the rumen is estimated from the fermentable organic-matter (FOM) content. The reference method to determine FOM is by in situ incubations of food in the rumen of fistulated animals. However, this is a cumbersome, time-consuming and expensive procedure. In practice, FOM is estimated from the digestible organic-matter content (DOMD) corrected for crude fat (Cfat), rumen escape protein (EP) and starch (ES) and the fermentation products (FP), using tabular values. Apart from being indirect, this approach takes little or no account of the variability within foods. Although an in vitro method based on rumen fluid may give a good prediction of in situ FOM for the majority of the foodstuffs (Cone et al., 1994), an enzymatic method is preferable for practical reasons. From the organic-matter (OM) degradation by a mixture of amylase and other carbohydrate degrading enzymes at 39°C during 0, 2, 6 and 24 h, Cone et al. (1996) obtained a moderate correlation with in situ FOM for 29 grass and grass silage samples (R2 = 0.88; residual s.d. = 37 g/kg dry matter (DM)) and 80 samples of concentrates, maize and clover (R2 = 0.77; residual s.d. = 42 g/kg DM). In this study a modified version of the enzymatic method of de Boever et al. (1994), normally used to predict the in vivo OM digestibility and energy value of concentrates, was tested for its reliability to predict in situ FOM of compound foods and forages.