The invention provides a binary sort tree-based local-bargaining spectrum allocation method. The method comprises the steps of dividing a scanned frequency spectrum into a plurality of signal channels at the initial moment, arranging the signal channels in an ascending sequence from small to large to construct a single-branch binary sort tree, allocating the signal channels to users based on the allocation algorithm to acquire an allocation matrix, updating users at each node of the binary sort tree according to the allocation matrix, updating the binary sort tree according to the variation of the scanned frequency spectrum during the next time period, and simultaneously reallocating the signal channels according to the variation of users that apply for signal channels. The above process is repeated until a command for stopping the frequency spectrum scanning operation is received. According to the technical scheme of the invention, the overall scanning result of frequency bands is in the form of a tree. In this way, the spectrum allocation is realized only through the inserting operation and the deleting operation on the binary sort tree. Therefore, the allocation time is saved. Under the circumstance of the lack of frequency spectrum resources, the spectrum utilization rate is increased and the satisfaction level of users is improved at the same time.