Assistive technology allows motor-impaired people to overcome limitations. Several myoelectric interfaces have been developed, however, there is no reported study employing information at a motor unit (MU) level for controlling purposes. Thus, we developed a facial myoelectric interface operating at the level of MU for controlling a computer screen cursor. Data were collected from 11 able-bodied and 1 tetraplegic subjects. Different from traditional approaches, there was no significant difference (p<0.05) in learning with respect to the level of difficulty, occurring evenly and faster. Information at MU level opens new possibilities for the development of fine control myoelectric interfaces.