Although the terminology and surface characterization of attitudes within the various health models varies greatly, each major health behavior model includes an evaluative component of (or attitude towards) the behavior, specifically explicit attitudes. Inherent limitations of explicit measures highlight the need for health researchers to incorporate implicit measures into their models of behavior, especially in the context of condom use. Moreover, the multidimensional nature of condom attitudes has been clearly demonstrated in explicit condom attitudes, and this perspective needs to be applied to implicit condom attitudes as well. Two studies were conducted, one aimed at confirming the hypothesized multidimensional structure of implicit and explicit condom attitudes, and examining the stability of two implicit attitude assessment methods. The second study examined the effects of health messages (one emphasizing the disease-prevention aspects of condoms, one highlighting eroticization of condom use, and a control message) on implicit and explicit attitudes, as well as longitudinally examined the effects of implicit and explicit attitudes on condom use. Results confirm the multidimensional and ambivalent nature of implicit condom associations, and that an implicit condom + disease association may be detrimental to condom use. Although the manipulation of message did not affect implicit associations as hoped, the data indicated the importance of an indirect effect of strong condom + sex associations on intentions to use condoms, which in turn predict behavior. Implications for theory and intervention are discussed.