By introducing a response-time measure in the field of comparative optimism, this study was designed to explore how people estimate risk to self and others depending on the evaluation order (self/other or other/self). Our results show the interdependency between self and other answers. Indeed, while response time for risk assessment for the self does not vary depending on the evaluation order, assessment for an average other takes longer in the other/self than in the self/other order. Thus, prior self-estimate appears to facilitate the construction of subsequent risk estimates for others. Our data indicate that the two evaluation orders do not correspond to strictly equivalent procedures; the other/self order requires extra cognitive effort and leads to greater comparative optimism levels. These findings should alert researchers that the measurement procedures, classically considered as interchangeable in the area of comparative optimism, actually generate different ways to construct the answers. Our results are discussed in terms of anchoring and asymmetry effects. (Contains 3 tables and 1 footnote.)