In most of the currently used protein evaluation systems for dairy cattle, duodenal supply of microbial protein is estimated from degraded organic matter. The degradation of organic matter in the rumen has been considered a feed characteristic. Therefore, individual differences between cows and their ruminal flora are not taken into account. In the current investigation milk odd and branched chain fatty acids, which are derived from the ruminal bacteria, are used to evaluate the extent of rumen degradability of organic matter. The experiment was set up as a 4 x 4 Latin square with four diets differing in source of starch in the concentrates (beet pulp, maize, wheat and potato). The four cows were rumen and duodenum fistulated. Flows for organic matter were measured, from which measured apparently rumen degraded organic matter (ARDOM(meas)) has been derived (min: 439 g/kg DM, max: 604g/kg DM). Milk samples were analyzed for odd and branched chain fatty acids. The ARDOM has also been calculated from equations described by Tamminga et al. (1994) (ARDOM(calc), g/kg DM): 476, 482, 476 and 494 for concentrates with beet pulp, maize, wheat and potato respectively. In an attempt to improve predictions of ARDOM, linear regression was performed to predict residuals (ARDOM(meas) - ARDOM(meas)) by OBCFA's (SPSS, version 11.0, backward procedure, significance of factor <0.10). The residuals were predicted by milk C17:1 cis and C17:0. The equation showed higher predictive power compared to ARDOM(calc) (R-2= 61 vs. R-2= 0). This was more clearly shown by examining the 95% confidence interval (C.I.) around linear regression (Neter, 1996) in which only 19% of the ARDOM(calc), were enclosed. From these results, it can be concluded that ARDOM prediction, based on feed values, can be improved by OBCFA's.