Although the importance of practice has been evidenced in early studies of cross-limb transfer, the association between cross-limb transfer and practice organization remains unknown. The two primary means of organizing practice are constant practice (CP) and variable practice (VP). When the same generalized motor program governs the motor responses, VP produces better transfer than CP. Thus, we hypothesized that VP would generate a higher cross-limb transfer level than CP. We assigned 40 participants to CP or VP groups and conducted an experiment consisting of three phases: pre-test, practice, and transfer test. At pre-test, all participants practiced eight trials of a sequence key-pressing task with the non-dominant hand (NDH). After the pre-test, all participants performed a practice phase of 72 trials with the dominant hand, but CP and VP groups underwent their different practice schedules (CP or VP) such that the CP group practiced a sequence key-pressing task in one sequence, and the VP group practiced four different sequences randomly. On the transfer test all partcipants completed eight trials with the NDH 24 hours after the practice phase. The CP group showed better performance than the VP group on the transfer test, and we concluded that the effects of practice organization in cross-limb transfer are distinct from intra- and inter-task transfer such that the specificity of practice explained the benefits of CG.