Multi-sensor information fusion technology can effectively improve the ability of radar to find and detect targets, and the prerequisite of successful multi-sensor information fusion is system error registration. Too large system error of radar will bring fuzziness and difficulty to track correlation and fusion, which makes the performance of fusion system degrade seriously. Aiming at the problem of multi-sensor error registration, this paper presents the solutions of time error registration and spatial error registration. By transforming the spatial error registration problem into a nonlinear optimization problem, the genetic algorithm is used to search for the approximate optimal solution of the problem. Through the real radar observation data and MATLAB simulation experiments, the effectiveness of genetic algorithm in solving the problem of multi-sensor error registration is verified.