The present study was an attempt to examine the role of cue category level (symptom‐contingent vs. hypothesized cues) and mode of stimulus presentation (in vivovs. imaginal) in exposure (flooding) therapy. Sixty‐four speech phobics were randomly assigned to one of eight conditions. Exposed to either one of the two cue category conditions, subjects received either two sessions ofin vivoexposure, two sessions of imaginal exposure, or two sessions consisting of either an imaginal followed by anin vivoexposure or vice versa. Results support the predictions of Levis&Hare (1977) that the inclusion of hypothesized cues is not necessary in the treatment of analogue phobias or simple clinical cases. These data also support and extend several previous findings thatin vivoexposure produces more rapid anxiety decrement than imaginal exposure. Finally, a variant of the usualin vivoprocedure is described which apparently is effective, yet does not require the assemblage of a large audience for eachin vivoexpos