Water and mud inrush is one of the typical geological disasters during tunnel construction in karst areas.In order to effectively evaluate the water and mud inrush risk and guarantee the safety of tunnel construction,An improved risk evaluation method was put forward which combines artificial bee colony algorithm and BP neural network(ABC-BP).First,according to geological factors in tunnel zone,unfavorable geology,formation lithology,ground water level,recharge area of surface water,attitude of rocks,crack in rock stratum,are selected as risk evaluation indexes of water and mud inrush,which constitutes the index system of water and mud inrush risk evaluation.Through collecting and collating information of typical water and mud inrush examples of karst tunnels,the expert scoring of each evaluation index is obtained based on the measured data.Meanwhile,Weight and threshold problem of BP neural network was transformed to the process of searching the best nectar for honey bees and the risk grade of water and mud inrush is judged by ABC-BP model.