The contextual interference (CI) research related to the Magill and Hall (1990) hypothesis concerning task characteristics was reviewed to reexamine the hypothesis. Through this process, describing the weaknesses of the hypothesis led to proposing a multiple factor approach to the generalizability of the CI effect. A modification to the Magill and Hall hypothesis was proposed considering the interaction between task characteristics and the amount of practice. The modified hypothesis holds that when task variations are controlled by different generalized motor programs (GMPs), the CI effect would be found in both early and later practice. In contrast, when task variations are controlled by the same GMP, the CI effect would be found only in later practice.