The focus of this article is on the temporal capacity of short-term visuomotor memory as reflected by changes in the time and frequency patterns of force output. In experiment 1, subjects produced continuous force output (isometric index finger flexion) to a target force level (from 5 to 75% of maximum voluntary contraction, MVC) displayed on a video monitor for 20 s. In the full visual feedback condition, visual feedback was displayed throughout each trial, while, for the visual feedback-withdrawal condition, visual feedback was occluded for the final 12 s of each trial. With visual feedback present, subjects matched their force output to the target force level for 20 s. When visual feedback was removed, participants continued to match the target force level for approximately 0.5-1.5 s; thereafter force output decayed exponentially. In line with this decay, short time-frequency analysis revealed a decrease in force intensity in the 0- to 5-Hz band. Force level did not influence the time before decay; however, greater forces led to larger decay. Experiment 2 assessed whether the force decay in experiment 1 was a property of visual or motor short-term memory by having participants set their own target force levels with no visual information provided throughout. In agreement with the findings of experiment 1, force output decayed, emphasizing the importance of a motor memory source. It is concluded that the 0.5- to 1.5-s time period represents a limit on the temporal capacity that precise visuomotor information is held in short-term memory.