Self-efficacy was analyzed as a potential moderator of implementation intention effects on goal attainment. Participants’ self-efficacy with respect to taking an analytic reasoning test (Advanced Progressive Matrices; Raven, 1976) was manipulated before they formed the goal to perform well. Next, all participants learned about double checking as an effective strategy to improve test performance, but only in the implementation intention condition did they put this strategy into an if–then plan. The analytic reasoning test was comprised of easy, medium–difficult, and difficult items. Implementation intentions advanced performance on difficult items when high self-efficacy had been established, but not when self-efficacy was low. The time participants spent solving the Raven items mediated this implementation intention effect on performance.