Guided by the research on cognitive performance tasks (e.g., Wood, Mento, & Locke, 1987), the potential moderating effects of skill and outcome difficulty upon the relationship between specific, difficult goals and physical task performance were examined from an attribute treatment approach. Overall, different measures of performance yielded several convergent findings. Using a sample of 184 laboratory participants, a test of the primary hypothesis failed to support the expected three-way interaction among skill, outcome difficulty and goal setting. Models containing skill and outcome difficulty were found to provide the most parsimonious explanation of variance in performance, regardless of whether assigned or personal goal were used as an index of motivation. After controlling for skill and outcome difficulty, goal commitment showed a weak positive relationship with task performance. Self-efficacy was not related to performance among participants pursuing specific, difficult goals. Potential reasons for the failure to obtain evidence of the predicted three-way interaction among skill, outcome difficulty, and goals are discussed in the context of limitations to the present study. Future research directions are suggested.