The evaluation and selection of weapon system is one of the most important decision issues for military logistic managers. Owing to vague concept frequently represented in decision data, a new group fuzzy multi-attribute decision-making approach is proposed to solve the weapon system optimal selection problem. In the proposed method, the experts' opinions are described by trapezoidal fuzzy numbers, and the fuzzy Delphi method is adopted to adjust each expert's opinion to achieve the consensus condition. By using the logarithmic least square method and trapezoidal fuzzy number arithmetic operations to estimate the normalized fuzzy priority weights, a new trapezoidal fuzzy analytic hierarchy process (AHP) is presented and consequently the consistency check of fuzzy judgment matrix is avoided. Finally, a practical example of selecting the best unmanned combat air vehicles (UCAV) demonstrated its engineering practicability and examined its effectiveness. The results show that the method is flexible and credible and provides references in intelligent battle command decision-making