We compared matching of finger forces under isometric conditions with matching of forces produced against a spring load (anisometric conditions) in twenty normal subjects. The instruction was to generate the same force in both hands holding a grip between thumb and index finger in each hand. Visual feedback indicating the target force and the actual force applied were presented for one (reference) hand only. The forces produced in each hand were measured continuously during matching trials. A special device provided the opportunity to change from isometric to anisometric force production. Matching was required under symmetric conditions, in which force was generated in both hands either isometrically or anisometrically, as well as under asymmetric conditions in which isometric force has to be matched to anisometric force or the reverse. Under symmetric conditions matching error was consistently smaller in anisometric than in isometric force production. However, the striking feature was a severe mismatch between hands when forces had to be produced differently. For most subjects, a force generated against the spring load in the reference hand was greatly overestimated by the matching hand working isometrically. For the reverse condition consistent underestimations were observed. This effect cannot be attributed to left/right differences or a simple confusion of subjects in the asymmetric tasks. Some of the factors confounded with the conditions of force production were ruled out as an explanation by additional experimental controls. The mismatch neither depends on signals related to different finger positions associated with target forces nor is it alleviated when differently produced forces are matched sequentially.(ABSTRACT TRUNCATED AT 250 WORDS)