Human limb movements are successfully modeled based on the assumption that the central nervous system controls the movements by maximizing movement smoothness. Movement smoothness is quantified by means of a time integral of squared jerk (jerk-cost), where jerk is defined as the rate of change in acceleration. This study was performed to investigate whether the control of human masticatory vertical jaw movements can also be explained by a minimum-jerk (maximum-smoothness) model. Based on the assumption that minimum-jerk models account for vertical jaw-opening and -closing movements during chewing, the actual time profile of the movement trajectory was simulated by the model. The simulated jerk-costs and peak velocities were compared with those obtained by actual measurements of jaw movements during chewing. Jerk-costs and peak velocities of the jaw movements during chewing were significantly correlated with those predicted by minimum-jerk models (P < 0.0001, r between 0.596 and 0.799). The minimum-jerk models predicted closing movement trajectories more accurately than opening movement trajectories (jaw opening, root-mean-square error = 1.19 mm; jaw closing, 0.52 mm, t = 4.375, P < 0.0001). The results indicated that the vertical jaw movement control during chewing was represented by the minimum-jerk control model and that the vertical jaw-closing movement is smoother than the opening movement during gum-chewing.