A method for predicting drowsiness is disclosed. By obtaining average heart beat rate values of a driver, and according to the characteristics of the heart beat rate values over a period of time, the method is utilized to determine whether the human being is going to sleep. The method comprises the following steps: detecting a heart beat rate of a driver calculating a curve of the heart beat rate average during a time interval of X minutes determining an accumulated length of duration during which the calculated linear regression slope values are smaller than the predetermined slope value Z determining whether the accumulated length of duration is greater than a time threshold T to generate a drowsiness detecting result and determining whether to raise an alarm based on the drowsiness detecting result.
