In most cases, it is very difficult to estimate the accuracy and reliability of different motion (optical flow) estimation algorithms. Most authors use well known image sequences to test their techniques and compare them with the work of others. Barron et al. \cite{barro94a} have reported experiments with 9 different algorithms where four such sequences are used. Most algorithms perform well for simple image sequences. In this work test sequences is designed for different situations: translation, rotation, occlusion, and multiple motion, with and without noise. We have used these sequences to test the algorithms presented in. Our findings give a more diverse view of the capabilities of the different approaches.