To enhance the risk management strategies for crashes in mountainous freeways, this study conducted a data mining using the mountainous freeway in Fuzhou as an example. The K-prototype algorithm was used to cluster crash types into three categories with a SSE of 6450.049. The three categories reflects different crash characteristics, which can be used to implement different strategies. C1 represents crashes with high death rates, primarily occurring in confluence zones and tunnels during holidays with high traffic flow. C2 represents crashes with high injury rates, occurring on bridges, long downhill slopes, and diversion zones during evening peak hours. C3 represents crashes with low severity but long rescue response times. Different risk management strategies such as visual guidance devices and recommended speed reminders, were proposed for the three types of crashes mentioned above. The result also showed that more attention should be paid to the peak hours and holiday time in critical areas to minimize the influence of mountainous freeway crashes.