Three experiments utilizing a 14-element arm movement sequence were designed to determine if reinstating the visual-spatial coordinates, which require movements to the same spatial locations utilized during acquisition, results in better effector transfer than reinstating the motor coordinates, which require the same pattern of homologous muscle activation. Results demonstrated better transfer when visual-spatial coordinates were reinstated than when motor coordinates where reinstated regardless of the amount of practice (1, 4, or 12 days; Experiments 1-3, respectively). Transfer (left to right and right to left) was symmetric when visual-spatial coordinates were reinstated but not when motor coordinates were reinstated. When motor coordinates were reinstated after 12 days of practice and vision occluded, transfer was better from right limb to left than vice versa. The data are also consistent with the notion that multiple codes (visual, spatial, and motor) are developed over practice, with each contributing to transfer performance when the respective coordinates are reinstated. Further, the results indicate a disruption of the linkage (concatenation) between subsequences when one or more coordinates are changed on the transfer test. (Contains 9 figures and 3 footnotes.)