This study investigated how movement error is evaluated and used to change feedforward commands, following a change in the environmental dynamics. In particular, we addressed the question of whether only position error information is used or whether information about the force field direction can also be used for rapid adaptation to changes in the environmental dynamics. Subjects learned to move in a position-dependent force field (PF) with a parabolic profile and the dynamics of a negative spring, which produced force lateral to the left of the target hand path. They adapted very rapidly, dramatically reducing lateral error after a single trial. Several times during training, the strength of the PF was unexpectedly doubled (PF2) for two trials. This again created a large leftward deviation, which was greatly reduced on the second PF2 trial, and an aftereffect when the force field subsequently returned to its original strength. The aftereffect was abolished if the second PF2 trial was replaced by an oppositely directed velocity-dependent force field (VF). During subsequent training in the VF, immediately after having adapted to the PF, subjects applied a force which assisted the force field for approximately 15 trials, indicating that they did not use information about the force field direction. We concluded that the central nervous system uses only the position error for updating the internal model of the environmental dynamics and modifying feedforward commands. Although this strategy is not necessarily optimal it may be the most reliable strategy for iterative improvement in performance.