The aim of this study was to test whether an Implicit Association Test (IAT) with self- and social anxiety-words is sensitive to differences in trait social anxiety, and to an experimental induction of social anxiety. This was performed in the context of a partial replication of a previous study, in which Mauss et al. (2004) compared high and low trait socially anxious individuals before and after a social anxiety induction (an impromptu speech). Mauss et al.'s findings were replicated; that is, (i) the social anxiety induction produced increases in self-rated anxiety, self-rated physiological responses, and actual physiological arousal; and (ii) higher trait social anxiety was associated with stronger self-rated anxiety and stronger self-rated physiological responses, but not with stronger actual physiological responses. In addition, the results showed higher IAT social anxiety scores, both (i) as a result of the social anxiety induction, and (ii) as a function of self-reported trait social anxiety. It is suggested that the IAT may be a useful method for the experimental study of automatic evaluational thought patterns.