Optimization selection of test points is an important step of testability design for complex equipment, so a Discrete Firefly Algorithm(DFA) used for solving optimization selection problem of test points was proposed. First of all, the mathematical model of optimization selection problem of test points was built, then discretization improvement was conducted on the traditional firefly algorithm, and the implementation steps of the DFA were given, later the effect of different attraction functions and binarization functions (sigmoid and tanh functions) on the result of the algorithms was also analyzed. Finally, The DFA was applied to five real systems with different sizes to verify the effectiveness, and the computational efficiency of DFA was compared with particle swarm optimization (PSO) and genetic algorithm (GA). In premise of complying with fault detection rate and fault isolation rate the system requires, optimal value of test cost for 5 systems from proposed DFA respectively reduced by 10.1% and 14.6% compared with PSO algorithm and GA algorithm. The experimental result shows: DFA can quickly converge to the global optimal solution of higher quality, and it can avoid trapping into local optimal solution, so it has very good application prospect to solve optimization solution problem of test points for large-scale complex equipment.