Adults who stutter (AWS) are reported to have increased levels of anxiety compared to adults who do not stutter (AWNS), particularly in social interactions (Kraaimaat, Vanryckeyham, & Dan-Baggen, 2002; Mahr & Torosian, 1999; Messenger, Onslow, Packman, & Menzies, 2004). However, the level of perceived anxiety in AWS according to specific speaking situations has not been critically evaluated. In addition, most studies addressing state anxiety (i.e., communication apprehension) are based on self-judgments (Craig, 1990; Craig, Hancook, Tran, & Craig, 2003; Ezrati-Vinacour & Levin, 2004; Gabel, Colcord, & Petrosino, 2002; Lincoln, Onslow, & Menzies, 1996; Messenger et al., 2004; Miller & Watson, 1992; Mulcahy, Hennessey, Beilby & Byrnes, 2008) which have not been verified using a physiological evaluation of anxiety. The present study sought to examine the relationship between a physiological measure of anxiety (i.e., cortisol) and perceptual judgments of communication apprehension across different speaking situations. Ten AWS aged between 19-62 years, and ten sex- and aged-matched AWNS provided salivary cortisol samples during distinct speaking situations across a one-week period. The speaking situations consisted of (1) speaking face-to-face with a friend, (2) speaking face-to-face with a single stranger, (3) speaking in front of a group of four strangers, and (4) speaking to a stranger on the telephone. Each participant also provided self-perception assessments of their perceived anxiety levels using an adaptation of the Speaking Task Response Scale (STRS; Bray & James, 2009) before and after each speaking situation. Results of the cortisol analysis revealed no statistical difference in cortisol levels across the four speaking situations between AWS and AWNS. A significant difference was found between self-perceived anxiety levels in the pre-speaking situation between AWS and AWNS. Speaking face-to-face with a friend was perceived by the AWS to result in the lowest level of anxiety compared to the remaining three situations. Correlational analyses revealed a significant relationship between cortisol levels and self-reported anxiety in the AWS group but no such relationship was evident for AWNS. On the basis of the combined results from the cortisol and self-perception analyses it can be concluded that AWS differ from AWNS in their communication apprehension, most notably in regard to speaking in any situation other than a familiar person (e.g., friend). This difference between AWS and AWNS is most evident in measures of self-perception, although it is likely there is an associated physiological contributing factor.