Fusing the grey bootstrap principle into Poisson process, the grey bootstrap Poisson method is proposed to forecast the variation process of reliability of the rolling bearing vibration performance. A small number of raw variation-intensity information presented by bearing vibration is extracted with the help of the counting process of time series in short time interval, a large number of generated variation-intensity information is simulated by means of bootstrap resampling from raw variation-intensity information, the estimated value of variation intensity is obtained by using the grey prediction model to process generated variation-intensity information, and the variation process of reliability of the bearing vibration performance is forecasted in time via the reliability function expressed as Poisson process. Experimental investigation on reliability of bearing vibration as a time series shows that variable states of performance reliability can be described truly and predicted values are in very good accordance with test values.