Summary

In order to reduce costs for farms using automatic milking systems (AMS) an attempt was made to estimate daily fat and protein contents from only one single milking sample of a test day. Over 91 % of the variation in daily protein content could be explained by linear regression on linear and squared effects of single test protein content. Investigated milking interval (MI) or mean milking interval (mMI), day time of sampling (DT), stage of lactation (DIM) and lactation number (Lact) did not explain a considerable additional amount of variation. However, for daily fat content R-values of the same models reached only values from 65.7 to 76.7%. Milking interval had the highest influence. Especially records with milk yields less than 2 kg were very questionable. Using the given data it was not possible to estimate daily fat content with a satisfying accuracy. Errors would even increased or multiplied when using the officially applied method for prediction of fat yield for the whole control period. Key words: automatic milking system, daily fat / protein content, varying milking interval / frequencie