QAM (Quadrature Amplitude Modulation) is broadly used in various communication systems thanks toits high power efficiency and bandwidth utilization.High- order QAM,because of its wide search range,may resultin high demapping complexity.In order to solve this problem,a novel concept of soft- decision contribution factor isintroduced to evaluate the contribution of reference constellation to the demapping,and a demapping algorithm ofhigh- order QAM within shrunk search range is proposed.Then,some boundary conditions of the algorithm are dis-cussed,and an adaptive selection scheme of search range is finally presented with the assistance of channel estima-tion.Simulated results indicate that (1) the proposed algorithm downsizes the search range without any sacrifice of BER performance; (2) 4096- QAM and 1024- QAM respectively shrink their search range to 12.7% and 19.7% ofthat of the whole constellation; and (3) as the proposed algorithm makes good tradeoff between performance anddemapping complexity.it has a good prospect of engineering application.