The effects of measurement method on levels of unrealistic optimism for developing diabetes were examined in 323 UK adults. A mixed factorial design comprised direct and indirect measures of unrealistic optimism as the within-subjects variable, and between-subjects manipulations of the order of direct versus indirect assessment, the order of self-other comparison, the number of response options used and the polarity of the ratings. More unrealistic optimism was obtained using the direct method and using unipolar scales, and/or with the self-other order in the indirect method. The results indicate that levels of unrealistic optimism depend on its assessment method.