OBJECTIVE: The goal of this study was to gain a broader perspective on social anxiety in the community than has been achieved by epidemiologic surveys to date. METHODS: The authors conducted a telephone survey of social anxiety among 526 randomly selected respondents in a medium- sized Canadian city. RESULTS: Sixty-one percent of the respondents reported being much or somewhat more anxious than other people in at least one of the seven social situations surveyed. Speaking to a large audience (i.e., public speaking) was the most frequently feared situation (endorsed by 55.0% of the respondents), followed by speaking to a small group of familiar people (24.9%), dealing with people in authority (23.3%), attending social gatherings (14.5%), speaking to strangers or meeting new people (13.7%), and eating (7.1%) or writing (5.1%) in front of others. When the threshold for caseness was systematically modified--by altering the required level of psychosocial interference or distress or by including or excluding subjects with pure public speaking phobia--the rate of "social anxiety syndrome" in the community varied from 1.9% to 18.7%; 7.1% was the prevalence when the criteria were set to conform with DSM-III-R. CONCLUSIONS: Social anxiety is common in the community, but precise delineation of the prevalence of "social phobia" depends heavily on where the diagnostic threshold is set. If DSM-III-R criteria had been applied in previous epidemiologic studies, it is likely that those studies would have documented prevalences of social phobia that are several times as high as the currently accepted rates.