Humans are able to adjust the accuracy of their movements to the demands posed by the task at hand. The variability in task execution due to the inherent noisiness of the neuromuscular system can be tuned to task demands by both feedforward (e.g. impedance modulation) and feedback mechanisms. In the present experiment, we investigated both mechanisms, using mechanical perturbations to estimate stiffness and damping as indices of impedance modulation and submovement scaling as an index of feedback driven corrections. Eight subjects tracked three differently sized targets (0.0135, 0.0270 and 0.0405 rad) moving at three different frequencies (0.20, 0.25 and 0.33 Hz). Movement variability decreased with both decreasing target size and movement frequency, while stiffness and damping increased with decreasing target size, independent of movement frequency. These results are consistent with the concept of neuromotor noise as proposed by Van Galen and Schomaker (1992, Hum Mov Sci, 11 (1-2):11-21) but challenge stochastic theories of motor control that do not account for impedance modulation and only partially for feedback control. Submovements during unperturbed cycles were quantified in terms of their gain, i.e. the slope between their duration and amplitude in the speed profile. Submovement gain decreased with decreasing movement frequency and increasing target size. The results were interpreted to imply that submovement gain is related to observed tracking errors and that those tracking errors are expressed in units of target size. We conclude that impedance and submovement gain modulation contribute additively to tracking accuracy.