Downhill mountain road is the accident prone section because of its complexity and variety. Drivers rely more on driving experience and it is very easy to cause traffic accidents due to negligence or failure of judgment. Traditional active safety systems, such as ABS, subjects to the driver's visual feedback, can’t fully guarantee the downhill driving safety in complex terrain environment. This study combines the characteristics of vehicle dynamics and geographic information, giving drivers safety speed in advance to reduce traffic accidents due to driver's miscarriage of justice or brake overheating and enhance the safety of vehicles in the downhill. Firstly, establish the commercial vehicle driving model, main brake and auxiliary brake model, and study the characteristics of the driving conditions under the change of slope length and slope. Gather the real-time information of the vehicle, such as vehicle weight, speed, gear and engine speed to calculate the vehicle braking torque. At the same time, the electronic map and GPS positioning is to calculate the gradient information. Finally, combine the brake efficiency and traffic rules, get the top speed of different downhill road and give drivers safety speed pre-warning. The result shows the pre-warning scheme has a great improvement on the safety of the vehicle in the length of 1 km to 1.5 km, the slope of 3% to 6% of the downhill. It can effectively reduce the traffic accidents.