A model is presented to investigate the optimal economic life cycle of grass leys with winter damage problems in northern Norway and to determine the threshold of winter damage before it is profitable to re-seed. A two-level hierarchic Markov process has been constructed using the MLHMP software. The model takes uncertainty concerning yield potential, damage estimation and weather dependent random fluctuations into account. A Kalman filter technique is used for updating of knowledge on yield potential and damage level. The application of the model is demonstrated using data from two commercial Norwegian farms. Since parameter estimates vary considerably among farms, it is concluded that decision support concerning optimal economic life cycle of grass leys should be done at farm level. The results also show the importance of using a flexible dynamic replacement strategy. Use of the model for a specific farm situation is illustrated.