Historically, anxiety has been a dominant subject in mainstream psychology but an incidental or even insignificant one in behavior analysis. We discuss several reasons for this discrepancy. We follow with a behavior-analytic conceptualization of anxiety that could just as easily be applied to emotion in general. Its primary points are (a) that language-able humans have an extraordinary capacity to derive relations between events and that it is a simple matter to show that neutral stimuli can acquire discriminative functions indirectly with no direct training; (b) that private events can readily acquire discriminative functions; (c) that anxiety disorders seem to occur with little apparent direct learning or that the amount of direct learning is extraordinarily out of proportion with the amount of responding; and (d) that the primary function of anxious behavior is experiential avoidance. We conclude that the most interesting aspects of anxiety disorders may occur as a function of derived rather than direct relations between public events and overt and private responses with avoidance functions. Implicit in this conclusion and explicit in the paper is the assertion that anxiety is a suitable subject for behavior-analytic study.