To deal with disordered data involving white noise and random missing points, a three-dimensional point cloud registration method based on factor analysis was proposed. First, the mathematical model of a point cloud was extended to an orthogonal factor model, transforming the point cloud registration problem into the model parameter solution problem. Then, a Gaussian mixture model was used to fit the point clouds, and the factor load matrix of an orthogonal factor model was obtained via the exponential moving average (EMA) method. Finally, the factor load matrix was used to perform point cloud registration. In a simulation experiment, the registration accuracy of the factor analysis algorithm for noisy point cloud data with missing points was found to be equal to that of the classical iterative closest point (ICP) algorithm, and 70% higher than that of the classical ICP algorithm. The factor analysis algorithm did not fall into local minima and could yield clear improvements in efficiency, registration accuracy, and stability.