The present study examined the kinematics of unrestricted reaches to prismatically-displaced targets. The kinematic analysis allowed us (1) to document how and where in the reach adjustments were made to compensate for the prismatic displacement. (2) to detail the changes that occur in the characteristics of reaches during the course of adaptation to the prisms, and (3) to look at the effects of providing information (or not) to the subject about the presence and nature of the prismatic distortion. The experiment differed from classic studies of prism adaptation in that subjects were permitted full visual feedback of their moving limb at all times, and entire reaching movements were recorded in addition to terminal errors. Experimental subjects were tested either with large-displacement prisms of the sort typically used in such experiments (20 diopters) or with small-displacement prisms (5 diopters) the properties of which went undetected in uninformed subjects. By using small displacements, it was possible to examine the process of visuomotor recalibration directly, free of contamination by "conscious" correction strategies. There were no differences in the terminal accuracies of the reaches made by subjects in any of the conditions. The availability of visual feedback allowed subjects to place their finger accurately on the target, despite the fact that in some cases their vision was displaced by as much as 11.4 degrees to the right. When the entire reach was examined, however, it was found that the amount of curvature in the path increased when large or small diopter prisms were unexpectedly introduced, with the subjects showing large deviations to the right. This rightward deviation was corrected in the final approach with a larger terminal correction. On some occasions, nonetheless, corrections were observed very early in the course of the reaching movement and appeared to be part of a natural process of trajectory fine-tuning. Uninformed subjects exposed to either large or small prismatic displacements also showed evidence of adaptation through an increased number of on-line corrections which compensated for a tendency to reach into the side of space opposite to the direction of the displacement (a "negative after-effect" in the path of the reach). Moreover, when questioned after the experiment, it became clear that uninformed subjects exposed to small prismatic displacements had apparently failed to detect any visual displacement whatsoever. Taken together, these results suggest that visuomotor recalibration can take place "automatically" without feedback from terminal errors and without the use of conscious strategies.(ABSTRACT TRUNCATED AT 400 WORDS)