ObjectivesAnxiety severity measures can be self‐report or observer‐rated. Although mostly these measures concur, they can diverge markedly. We examined concordance between two anxiety scales: the observer‐rated Brief Anxiety Scale (BAS) and the self‐report Brief Symptom Inventory 12‐item version (BSI‐12), and described associations between patient characteristics and discordance.DesignThe study used an observational design, using prospective data from 2,007 outpatients with DSM‐IV‐TR panic disorder with or without agoraphobia, agoraphobia without panic, social phobia, and/or generalized anxiety disorder.MethodsOverall agreement was described using Pearson's product–moment correlation coefficient. Associations between patient characteristics and discordance (defined as |Z‐BAS‐Z‐BSI‐12| ≥ 1) were evaluated with univariable and multivariable multinomial logistic regression analyses.ResultsOverall correlation between BAS and BSI‐12 was positive and strong (r = .59). Discordance occurred in 24.8% of patients ([Z‐BAS ≥ Z‐BSI‐12 + 1] = 12.2%; [Z‐BAS ≤ Z‐BSI‐12 − 1] = 12.6%). Patients with higher observed than self‐reported anxiety severity did not differ from concordant patients. Patients with lower observed than self‐reported anxiety severity more often had panic disorder, less often had social phobia, and had higher scores on cluster B and C personality characteristics than concordant patients. Lower observed than self‐reported anxiety severity was best predicted by panic disorder, social phobia, and affective lability.ConclusionsResults demonstrate that the use of a single source of information gives a one‐sided view of pathology. A multimethod approach is highly preferable, as this allows for assessment across different domains and through multiple sources of information, and as such, provides clinicians with vital information.Practitioner points When assessing anxiety severity, the use of self‐report measures provides additional information to observer‐rated measures. In patients who have strong cluster B and C personality traits, anxiety severity might be overlooked, even by trained observers. The use of a multimethod assessment strategy is preferable in anxiety severity assessment.