The aim of this study was to identify a set of feeding behaviour and activity related variables that could potentially detect a shortage of available feed for the individual cow on pasture. A group of lactating cows was offered 100% of their intake capacity as herbage allowance throughout a 10-week experimental period, while another group was offered 60% of their intake allowance, either for a two week or six week period in springtime. Each cow was equipped with an automated noseband sensor. The data were analysed by using a binomial generalized lineal model (GLM). The GLM was examined for the classification of full or restricted herbage allowance as a function of a previously identified set of characteristics. The model was further refined by including additional characteristics, which achieved higher prediction performance. For the combined data, the refined model achieved 77% accuracy, 75% sensitivity, 78% specificity and F-score 0.76 towards a decision support system for grass utilisation in pasture based milk production.