Measured the short-term retention of force responses in undergraduate volunteers. The dependent variables were the absolute and algebraic errors made by S in attempting to reproduce a criterion force during recall trials. In Exp. I (N = 20), retention was measured over 5 unfilled intervals ranging 4-6 sec. Forgetting, i.e., an increase in errors, was not found. Exp. II (N = 24) was a partial replication of Exp. I, except that during 1/2 of the retention intervals, S counted backwards. The result was a decrease in error over a 30-sec retention interval for both filled and unfilled conditions. Significantly larger errors were associated with the filled condition. In Exp. III (N = 24), the recall response shifted toward the relative magnitude of an interpolated force to the criterion force. The increase in error associated with filled retention intervals was successfully replicated. For Exp. IV (N = 27), successive repetitions of the criterion force prior to the recall trial produced an increase, not decrease, in error at recall. All 4 experiments were characterized by overshooting response sets (positive algebraic errors) at recall. Detailed comparisons with available data from earlier motor short-term memory (STM) studies indicate a set of consistent findings across the earlier and present studies, although apparent differences in memory functions were obtained. Consistencies were, with respect to the directional shifts, occurring in the algebraic error scores as a function of various independent variables. The different memory functions 200 215 231 231 231 296 379 413 43823142334 different response sets with these algebraic error shifts. A dual process theory of motor STM, incorporating decay and interference features, is advanced to account for the set of findings, and similarities with dual process theories of verbal STM are noted. (28 ref.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)