This paper is devoted to the behavior of human arms during pointing movements. Several assumptions have already been made about the planning of such motions. None of these assumptions is able, up to now, to explain certain nonintuitive dynamic phenomena, in particular certain asymmetries in the motion and certain time intervals of inactivity of the muscles. In this paper, we propose an assumption explaining all these phenomena. Two strong points in this work are the following. First, our assumption is that human beings minimize a certain criterion that physically makes sense, namely, a compromise between the absolute work of external forces and a comfort term. Second, our conclusions do not rely on any numerical experiment and are completely justified mathematically (i.e., without any argument from simulation or “experimental mathematics,” such arguments being usually considered as acceptable in neurobiology). Also, the conclusion that total inactivity holds during some time subintervals of the movement is shown to be a stable property (in our model).