It is commonly accepted that the modern CE instruments equipped with efficient cooling system enable accurate determination of electrophoretic or electroosmotic mobilities. It is also often assumed that velocity of migration in a given buffer is constant throughout the capillary length. It is simultaneously neglected that the noncooled parts of capillary produce extensive Joule heating leading to an axial electric field distortion, which contributes to a difference between the effective and nominal electric field potentials and between velocities in the cooled and noncooled parts of capillary. This simplification introduces systematic errors, which so far were however not investigated experimentally. There was also no method proposed for their elimination. We show a simple and fast method allowing for estimation and elimination of these errors that is based on combination of a long-end and short-end injections. We use it to study the effects caused by variation of temperature, electric field, capillary length, and pH.