In a cognitive model, Clark and Wells (1995) proposed that individuals with social phobia use bodily information to construct a distorted impression of their observable selves. In this study, we conducted a preliminary test of this hypothesis by manipulating the provision of heart rate information in four groups of participants scoring high or low on social-evaluative anxiety (SEA), and examined the effects of this manipulation on anxiety, perspective taking, and social performance. Comparisons were made between two high SEA and two low SEA groups who received information suggesting their heart rate had increased, or no information, prior to a threatening social interaction task. The results showed that high SEA individuals receiving information concerning an increase in heart rate reported significantly greater anxiety, negative social performance, and greater proportion of observer perspectives than high SEA individuals receiving no such information. This bodily information, however, did not affect participants' heart rates. The results provide support for Clark and Wells's model.