This paper studies and analyses the function and influence of parameter α,β,ρ in the three models of ant algorithm theoretically, taking TSP as an example. The computational results of oliver 30 city shows that the analysis on the three parameters is rational. Furthermore, we study the optimum configuration of the parameters. To improve the efficiency of the algorithm, we present to amend the strategy of choice used to decide the next city and propose a new method and a group of optimum parameters. Experimental results indicate that the change is practical and valuable.