According to the Challenge Point hypothesis (Guadagnoli & Lee, 2004), learning is optimised by balancing the difficulty of the skill, the experience of the learner and the conditions of practice (e.g., level of contextual interference). A specific prediction of the hypothesis is that highly skilled individuals should benefit from practicing under low contextual interference provided the task is sufficiently difficult. An experiment with sixteen highly skilled golfers (handicap ≤4) practicing the flop shot under low or high contextual interference conditions supported this prediction. To optimise learning, conditions of practice should be modified based on task difficulty.