鈥� Accurate assessment of "subdisorder" anxiety symptoms, ie, anxiety symptoms central to DSM-III-R鈥攄iagnosed anxiety disorders but not meeting disorder criteria because of insufficient frequency, duration, or accompanying subjective distress or impairment, may be critical to case identification in genetic, epidemiologic, and high-risk studies. However, concerns that the mild and often transient nature of these phenomena will foster unreliability have discouraged their use. We assessed the test-retest reliability of "subdisorder" anxiety symptoms in 104 outpatients with anxiety. Good to excellent agreement was found for lifetime occurrence of any panic attack, the spontaneous and situationally predisposed subtypes of panic, and five nonsocial irrational fears (public transportation, driving oneself, crowds, situations associated with death [eg, dead bodies and funerals], and cats and dogs). Four social and three additional nonsocial fears were considered to have adequate reliability. However, agreement on stimulus-bound panic, "near" panic attacks, persistent generalized anxiety, and the remaining nine nonsocial and six social irrational fears was only fair to poor. The major source of unreliability was variation in information reported by the subject to the rater.