This paper presents the basic principle and experimenting of a novel method called tree biomechanical optimization. The uniqueness of search strategies of this method is based on biomechanics phenomena exist in trees. Whereby, the mass of growth in a tree is used to perturb the solution, in addition, the diversification is performed by using bending moment constraints. Furthermore, deviation of solution from the best to the worst is also used to enhance search process. Finally, experiments are carried out on quadratic assignment problems which are known to be NP-hard, and then, comparative analysis is performed between the results obtained by the proposed method and its rivals. Consequently, the performance of the new method is better or comparable to the other state-of-the-arts algorithms.