Explicit memory for threat and non-threat stimuli was examined in Iranian adolescents aged between 11 to 18 years. Clinically anxious (n=29), subclinically anxious (n=27), and normal controls (n=29) were matched for age, gender and academic performance. Participants asked to encode the stimuli words including threat, depression, positive and neutral words, presented on a screen one after another. Content-specificity hypothesis predicts anxious individuals should show a memory bias favoring anxiety information. However, Williams et al's theory does not predict such a bias. The present study conducted to examine the two different theories.In sum, the results support the Williams et al's theory. Although, anxious patients compared to the sub-clinically anxious individuals, show a memory bias towards threat words. The difference between the two anxious groups and theoretical implications with regard 10the two theories are discussed.