BACKGROUND: Preliminary evidence suggests that impairment of social performance in socially anxious individuals may be specific to selective aspects of performance and be more pronounced in females. This evidence is based primarily on contrasting results from studies using all-male or all-female samples or that differ in type of social behaviour assessed. However, methodological differences (e.g. statistical power, participant population) across these studies means it is difficult to determine whether behavioural or gender-specific effects are genuine or artefactual. The current study examined whether the link between social anxiety and social behaviour was dependent upon gender and the behavioural dimension assessed within the same study under methodologically homogenous conditions. METHODS: Ninety-three university students (45 males, 48 females) with a mean age of 25.6 years and varying in their level of social anxiety underwent an interaction and a speech task. The speech task involved giving a brief impromptu presentation in front of a small group of three people, while the interaction task involved "getting to know" an opposite-sex confederate. Independent raters assessed social performance on 5 key dimensions from Fydrich's Social Performance Rating Scale. RESULTS: Regression analysis revealed a significant moderate association of social anxiety with behavioral discomfort (e.g., fidgeting, trembling) for interaction and speech tasks, but no association with other performance dimensions (e.g., verbal fluency, quality of verbal expression). No sex differences were found. CONCLUSIONS: These results suggest that the impairing effects of social anxiety within the non-clinical range may exacerbate overt behavioral agitation during high demand social challenges but have little impact on other observable aspects of performance quality.