The 3-D dynamic fire evacuation strategy based on the improved ant colony algorithm was proposed in this paper to solve the issue of emergency evacuation when fires occur in complex buildings. Firstly, the grid method was adopted to establish 3-D environment model for the designed building. Then, by using the improved bio-heuristic ant colony algorithm, evacuation routes were automatically mapped out to obtain the shortest evac- uation route. Finally, the best evacuation route was planned according to the dynamic diffusion characteristics of fire points. This approach not only improved the ant colony algorithm model, but also added the target heuristic function and fallback strategy, which, to a certain degree, shortens the time of path searching and improves the efficiency of the algorithm. The simulation results show that this strategy is effective in terms of 3-D fire evacua- tion.