The aim of the present study was to investigate memory effects, force accuracy, and variability during constant isometric force at different force levels, using auditory biofeedback. Two types of transition trials were used: a biofeedback-no biofeedback transition trial and a no biofeedback-biofeedback transition trial. The auditory biofeedback produced a low- or high-pitched sound when participants produced an isometric force lower or higher than required, respectively. To achieve this goal, 16 participants were asked to produce and maintain two different isometric forces (30$±$5% and 90N$±$5%) during 25s. Constant error and standard deviation of the isometric force were calculated. While accuracy and variability of the isometric force varied according to the transition trial, a drift of the force appeared in the no biofeedback condition. This result suggested that the degradation of information about force output in the no biofeedback condition was provided by a leaky memory buffer which was mainly dependent on the sense of effort. Because this drift remained constant whatever the transition used, this memory buffer seemed to be independent of short-term memory processes. ; Comment: Human Movement Science (2011) epub ahead of print