Objectives: To examine the prevalence of social anxiety disorders (SAD) with (DSM-IV) and without (DSM-5) the person's own assessment that the fear was unreasonable, in a population sample of older adults. Further, to determine whether clinical and sociodemographic correlates of SAD differ depending on the criteria applied. Design: Cross-sectional. Setting: General population in Gothenburg, Sweden. Participants: A random population-based sample of 75- and 85-year olds (N = 1200) without dementia. Measurements: Psychiatric research nurses carried out a semi-structured psychiatric examination including the Comprehensive Psychopathological Rating Scale. DSM-IV SAD was diagnosed with the Mini International Neuropsychiatric Interview. SAD was diagnosed according to DSM-IV and DSM-5 criteria. The 6-month duration criterion in DSM-5 was not applied because of lack of information. Other assessments included the Global Assessment of Functioning (GAF), the Brief Scale for Anxiety (BSA), and the Montgomery Asberg Depression Rating Scale (MADRS). Results: The 1-month prevalence of SAD was 2.5% (N = 30) when the unreasonable fear criterion was defined in accordance with DSM-IV and 5.1% (N = 61) when the DSM-5 criterion was applied. Clinical correlates (GAF, MADRS, and BSA) were worse in SAD cases identified by either procedure compared with all others, and ratings for those reporting unreasonable fear suggested greater (albeit nonsignificant) overall psychopathology. Conclusions: Shifting the judgment of how reasonable the fear was, from the individual to the clinician, doubled the prevalence of SAD. This indicates that the DSM-5 version might increase prevalence rates of SAD in the general population. Further studies strictly applying all DSM-5 criteria are needed in order to confirm these findings.