Subjects first adjusted 20 vertical lines to be equal in length to the horizontal line in small or large inverted-T figures and were then given 10 additional trials on the other size figure. Vertical lines were made significantly shorter than the horizontal line but accuracy gradually improved and performance was not disrupted by the switch in figure size. By Trial 30, adjusted vertical lines did not differ from the length of the horizontal line. Other subjects were first tested on the small inverted-T figure and then asked to produce 1-in. lines in the vertical plane and in the horizontal plane. These subjects also improved on the adjustment task but, on the production task, were no more accurate than control subjects who were not given practice on the adjustment task. Thus, the magnitude of the Horizontal-Vertical Illusion decreases with practice alone and this practice effect readily transfers to figures of other sizes but not to an alternate form of the illusion. The potential factors responsible for the improvement and transfer are discussed.