Many behaviors are performed less frequently than intended because they require knowledge and skill to overcome behavioral barriers. This experiment tested effects of two factors, direct experience and message frame, that were hypothesized to affect men's intention to perform the testicle self-exam for cancer (TSE) and their actual exam performance, in part by affecting their knowledge and beliefs about overcoming TSE performance barriers. Men's experience performing the TSE on a life-like model and the frame (negative, positive, or neutral) of the recommendation promoting the exam were manipulated factorially. Consistent with prediction, men who practiced (vs. did not practice) the self-exam: (a) endorsed stronger beliefs about behavior-specific knowledge (e.g., ability to imagine a lump), (b) held more positive intention, and (c) translated that intention more consistently into self-reported action, in part because their behavior-specific knowledge increased their tendency to act on their intention. Consistent with action phase theory (Gollwitzer, 1990), intention related more strongly to behavior-specific beliefs than to general, long-term beliefs, and more strongly than attitude related to behavior-specific beliefs. These findings have practical and theoretical implications for promoting healthful and other intended behaviors.'We thank Shelly Chaiken for providing BSE brochures containing the framing manipulation used in Meyerowitz and Chaiken's (1987) experiment. We also thank Roy Lilly for data analytic advice, for their help collecting or managing data. We are grateful to Suzi Aberasturi and Linda Kern for helping analyze the data. We also thank two anonymous reviewers for their helpful comments on an earlier version of this article.