Subjects classified according to degree of hand preference were tested with both hands on two tasks of controlled movement. Strongly lateralized subjects (both left- and right-handers) showed greater skill with their better hand than ambilaterals on a visually controlled aiming test (faster speed and equal errors), but there were no marked differences between groups with the other hand. Between-hand differences were also greater in lateralized subjects. On a ballistic rhythmical tapping test, no marked differences in speed were found, but ambilaterals made slightly fewer errors with the better hand. It is argued (i) that for ballistic movements the hands are equipotential, and skill is a direct function of practice, (ii) that the essential dexterity difference between the preferred and non-preferred hands is in the sensory or feedback control of movements rather than in motor function per se, and (iii) ambilaterals, especially those with very mixed preferences, have virtually two non-preferred hands in continuously controlled movement tasks, and may therefore rely on ballistic movements more than do pronounced sinistrals and dextrals.