The Prototype/Willingness Model (Gibbons, Gerrard, & Lane, 2003) posits that there are two pathways to predicting health-related risk behavior, one that is sensitive to situational changes (willingness to engage in behavior), and one that is relatively stable (intention to engage in behavior). This model makes two assumptions: (1) that individuals compare themselves to mental representations (or prototypes) of the typical risk-taker, affecting willingness to engage in risk behavior; and (2) because it depends on social comparison, willingness is more easily changed than intention. These assumptions were tested in three studies that looked at perceptions of the prototypical drinker. In Study 1, participants with favorable drinker prototypes reported liking a partner who drank more than those with unfavorable images, whereas the opposite was true for participants who learned about non-drinking partners. Study 2 expanded on Study 1 and found that participants who were similar to a drinking partner reported a positive association between prototype favorability and willingness to drink, whereas those who were dissimilar reported a negative relationship. Similarity was interpreted as indicative of interest in social comparison, so in Study 3 participants were explicitly instructed to socially compare with a drinking partner. Relative to a control group, male participants reported a stronger positive association between prototype favorability and willingness to drink, supporting the contention that social comparison increases the effect of prototypes on willingness to drink. This comparison effect did not occur for participants whose intention to drink was measured, supporting the contention that willingness and intention are separate cognitions about risk behavior.