The purpose of this study was to investigate memory of threatening and non-threatening information among adolescents. Specifically, the study tested the prediction of cognitive theories of anxiety that anxious and non-anxious individuals process threatening information differently. High school students (n = 187) from a moderately sized Midwestern city were screened for anxiety using self-report measures. Students (n = 18) who obtained scores above a designated cut-off, and a contrast group matched on gender and grade (n = 18), completed implicit and explicit memory tasks along with a second administration of the self-report measures. Subsequently, adolescents were divided into anxious (n = 8) and mixed (anxiety and depression; n = 10) groups. These groups produced more threat-related words on a word-stem completion task than the contrast group. However, their ability to reproduce the primed (i.e., previously studied) threat words did not differ from the contrast group. The mixed group identified more threat words on a word recognition task than either the anxious or contrast groups. When recognition was adjusted for guessing, no group differences existed. Results did not support the hypothesized implicit memory bias for threatening material among students experiencing symptoms of anxiety. However, consistent with predictions, no or explicit memory bias was found. Implications of memory methodologies in the study of anxiety are discussed. (Contains 4 footnotes and 4 tables).