A low complexity sub-optimal Automatic Sphere Decoding (ASD) algorithm is proposed. Aiming at the large size of the border node list in conventional ASD, the proposed method decreases the complexity of decoding efficiently through fixing the size of the border node list; and compensates the corresponding Fanolike metric bias which gets from simulation to assure the performance. It verifies the new algorithm through the simulation. And the simulation’s results demonstrate that compared with the conventional ASD, the complexity of decoding reduces greatly at the cost of less penalty on the performance, and it adapts for hardware implementation well.