The main contributions are two-fold: Firstly, some theoretical analyses are carried out on trinocular rectification, including the relationship among the three rectified images and their three fundamental matrices, and an geometric interpretation of the 6 free parameters involved in the rectification process. Such results could be used as a theoretical guide to reduce the induced projective distortion. Secondly, under the RANSAC (random sampling consensus) paradigm, a robust trinocular rectification algorithm is proposed. Unlike the traditional ones where only the fundamental matrices are used to rectify images, this algorithm instead uses directly corresponding points for the rectification. The main advantage of this point-based approach is that on one hand, the computation of fundamental matrices is usually prone to noise; on the other hand, good fundamental matrices do not necessarily always produce good rectified images because the two processes have different evaluation criteria. Extensive simulation and experiments with real images show that the proposed rectification technique is resistant to noise as well as to outliers of the corresponding points, and fairly good rectification results can be obtained.