Background: The efficacy of the skin self-examination (SSE) to detect artificial changes in the size of nevi has not been evaluated in a controlled setting. Objective: Our purpose was to determine the sensitivity and specificity of the SSE in detecting artificial changes in mole size in patients at high risk for melanoma. Methods: In a single-center, single-blinded cross-over study, patients who had been performing the SSE confidently for at least 1 year examined their backs after the diameter of an existing mole was increased artificially in random order by 0, 2, or 4 mm. Results: The specificity of the SSE was 62% (95% confidence interval [CI], 53%-72%) (N = 103). The sensitivity of the 2 mm change was 58% (95% CI, 49%-68%) and that of the 4 mm change was 75% (95% CI, 66%-83%). SSE performance was not related to perceived risk, number of moles, gender, age, or frequency of self-examination. Conclusion: Even in our highly motivated and selected group of high-risk patients, 25% could not detect an obvious increase in the diameter of an existing nevus, whereas 38% incorrectly identified a change when none was made. The SSE is only a moderately effective tool for the detection of acute, large, changes in mole size. The usefulness of the SSE in detection of new lesions or changes in existing lesions is likely due to a combination of factors or due to factors other than size, such as color, border irregularity, and texture, among others. (J Am Acad Dermatol 2000;42:754-9.)