Social anxiety is widely thought to be positively associated with alcohol use. However, these studies rely primarily on self-report of drinking behavior. This research aimed to further explore this research by assessing blood alcohol concentration (BAC), a physiological measure of alcohol intoxication, as the dependent measure in naturalistic settings (i.e., fraternity parties). Results from Study 1 suggest a weak relationship between self-reported anxiety levels and BAC. Study 2 was based on Burke and Stephen's (1999) proposed social cognitive model in which alcohol expectancies and drink-refusal self-efficacy act as moderators of the relationship between dispositional social anxiety and alcohol consumption. Analyses (n=86) did not support a relationship between social anxiety and alcohol use, nor the moderating effects of alcohol expectancies and drink-refusal self-efficacy. Implications for this type of research and possible future directions are discussed.