In this paper, a fast block motion estimation algorithm that not only accelerates the process of the motion estimation but also maintains the same accuracy as that of the full search algorithm is proposed. In modern CPUs, the single instruction multiple data (SIMD) technique is commonly used to provide an execution speedup of up to eight fold. The proposed algorithm satisfies the requirements of (i) the easy computation of the lower bounds of the block matching error, and (ii) their storage in a contiguous memory space for its SIMD implementation. Simulation results show that the proposed algorithm is more than 40 times faster than the full search algorithm while keeping the accuracy of the latter.