Abstract: This study considered a model of an Ant Colony Optimization (ACO) algorithm for the general combinatorial optimization problem. The model proved that it can converge to one of the optima if only this optimum is allowed to update the pheromone model and that it can not converge to any of the optima if two or more optima are allowed. The iteration complexity of the model can be computed easily. And then a lower bound of time complexity of a real ACO algorithm for the general combinatorial optimization problem can be obtained.