This study examined transfer when different serial positions are changed in a task of tracking a sequence of light stimuli. Thirty subjects were divided into three groups, and all tracked a serial pattern of six movements for 20 acquisition trials. Then, on 20 more transfer trials, the last two movements were reversed for Group I, the second two movements reversed for Group II, and the first two movements reversed for the Group III. Performance during the transfer trials improved over performance during the acquisition trials for Groups I and II but not Group III. Thus, there appeared to be positive transfer when the last and middle parts of the serial pattern were changed. However, there was no positive transfer when the first part of the serial pattern was changed. This indicated a contextual interference effect dependent upon serial position in the performance of a serial tracking task.