Psychometric scales for the assessment of anxiety disorders have not been validated in primary care settings. We undertook a study to validate two brief, self-administered anxiety scales, Sheehan's Patient-Rated Anxiety Scale and Beck's Cognition Checklist, in a university primary care clinic. The two scales were compared with a diagnostic standard, the Structured Clinical Interview for the revised third edition of the Diagnostic and Statistical Manual (SCID). Of 87 randomly selected patients, 25 had an anxiety disorder by the SCID. Anxious patients scored statistically higher on both scales than nonanxious persons on both the Beck and Sheehan scales. Based on receiver operating characteristics curve analysis, we found that these two scales correctly classified 72% to 77% of anxious patients. Of the two scales, the Sheehan scale performed better, with a specificity of 94% and positive predictive value of 75%. Our results suggest that these instruments may be useful in the assessment of the anxiety disorders in medical populations.