The present invention provides a method for improving fuel economy in hybrid vehicles. More precisely, the method of the present invention includes a selectable "learn mode" in which the vehicle can learn a particular driving schedule. After a particular driving schedule is learned, the schedule can be recalled to optimize fuel economy on future trips following the same route. By recalling a learned schedule, the system knows in advance what energy level will be needed and can proactively charge the hybrid vehicle's battery to an appropriate level.