Abstract: In recent years, there are more and more applications of computational fluid dynamics (CFD) in bridge wind-engineering, as a result of the rapid development of the floating-point calculation speed of computer hardware. CFD has enormous advantage over the wind-tunnel experiments in terms of the ability of dealing with complex cases, time-consumption, and economic costs. Such advantage, however, must be based on an assumption that the CFD solutions are always of practical accuracy. In this paper, the derivation of the turbulence problem in CFD numerical simulation, the essences of various turbulence models, and limitations and applicability of these models have been summarized. Furthermore, applications of CFD in bridge aerodynamic load simulations worldwide have been investigated, and statistical analysis has been conducted in regard to the simulation errors of the CFD results in comparison with experimental results, presented in terms of aerostatic and aero-elastic classes. The statistics show that, for aerostatic problems, simulation errors of CFD locate generally in the range from 10% to 30%; for aero-elastic problems, however, the errors locate generally from 20% up to as large as 300%. Therefore, the current CFD simulations, when independent on wind tunnel experiments, are incapable of providing reliable assessment of bridge aero-elasticity. To take full advantage of the CFD simulations, it is necessary to strengthen basic investigations regarding the applicability of turbulence models to various bridge aerodynamic issues, in combination with recognition of turbulence theories.