Voluntary activation of muscle is commonly quantified by comparison of the extra force added by motor nerve stimulation during a contraction [superimposed twitch (SIT)] with that produced at rest by the same stimulus (resting twitch). An inability to achieve 100% voluntary activation implies that failure to produce maximal force output from the muscle must have occurred at a site at or above the level of the motoneurons. We have used cortical stimulation to quantify voluntary activation. Here, incomplete activation implies a failure at or above the level of motor cortical output. With cortical stimulation, it is inappropriate to compare extra force evoked during a contraction with the twitch evoked in resting muscle because motor cortical and spinal cord excitability both increase with activity. However, an appropriate "resting twitch" can be estimated. We previously estimated its amplitude by extrapolation of the linear relation between SIT amplitude and voluntary torque calculated from 35 contractions of >50% maximum (Todd G, Taylor JL, and Gandevia SC. J Physiol 551: 661-671, 2003). In this study, we improved the utility of this method to enable evaluation of voluntary activation when it may be changing over time, such as during the development of fatigue, or in patients who may be unable to perform large numbers of contractions. We have reduced the number of contractions required to only three. Estimation of the resting twitch from three contractions was reliable over time with low variability. Furthermore, its reliability and variability were similar to the resting twitch estimated from 30 contractions and to that evoked by conventional motor nerve stimulation.