Every year, more than 100,000 automobile crashes are caused by driver drowsiness. Various technologies have been developed to address this issue, including vehicle-based measurements, behavior change detection, and physiological signal analysis. Both vehicle-based measurements and behavior change detection require bulky components. They also identify the driver's drowsiness too late for effective accident prevention. The physiological signal changes in an early stage and can be used to detect the on-set of driver drowsiness. In this paper, the development of a wearable drowsiness detection system is introduced. This system measures the electrooculography (EOG) signal; transmits the signal to a smartphone wirelessly; and could alarm the driver based on a prediction algorithm that can estimate 0.5-second-ahead EOG signal behavior. This system is compact, comfortable, and cost effective. The 0.5-second-ahead estimation capability provides the critical time for a driver to correct the behavior, and ultimately saves lives.