The invention relates to a method of predicting when a grease in a grease-lubricated rolling element bearing (10) will reach an unacceptable condition. According to the invention, a time series of a parameter is measured and recorded, whereby the parameter is directly or indirectly indicative of a degree of breakdown of a lubricant film separating the rolling contacts. With the aid of processing means (22), the time series is analyzed, a dynamic model is constructed that predicts the evolution of parameter values, and the model is used to extrapolate the time series. A time at which a predefined limit will be exceeded is then determined from the extrapolated time series, whereby the predefined limit is representative of unacceptable grease condition. The method according to the invention can be used to predict a relubrication interval for an individual grease-lubricated bearing.