According to the model proposed by Stampfl and Levis (1967), implosion reduces fear by evoking anxiety in the presence of cues associated with the phobic object, but in the absence of primary reinforcement. An essential requirement for rapid therapeutic effect is that the patient be made to experience intense affect, since “the greater the non-reinforced anxiety experienced, the greater the extinction of fear” (Hogan and Kirchner, 1967). Thus, the patient is repeatedly presented with fear-evoking stimuli and kept at a high level of arousal until ‘a significant diminution in anxiety has resulted’ (Stampfl and Levis, 1967). This model assumes that this ‘significant diminution in anxiety’ is equivalent to extinction of the fear response. Only a handful of studies of implosion, using human subjects, have attempted to relate the pattern and magnitude of fear during therapy to outcome measures. Hogan and Kirchner (1967) measured heart rate continuously throughout implosion. Compared to a no-treatment control procedure, implosion generated significantly greater heart rate arousal during therapy and greater behavioral improvement at assessment, suggesting support for the extinction model. In contrast, Watson and Marks (1971) found no correlation between overt anxiety during flooding and measures of outcome. Stern and Marks (1973) found ‘little skin conductance activity or tachycardia’ during flooding in fantasy, rather than the high level of arousal required by Stampfl's model. The present study attempted a partial test of some of the model's assumptions within the context of an outcome study of implosion therapy by tape recording. Kirchner and Hogan (1966) have claimed that a single, brief session of implosion therapy by tape recording can dramatically reduce phobic avoidance behavior. However, Hodgson and Rachman (1970) and Fazio (1970) failed to replicate these results. Findings from these and other studies of implosion are difficult to interpret because of the absence of objective measures of fear during the therapy session. In the absence of such measures, it is difficult to judge whether the implosive procedures succeeded in creating the pattern of arousal demanded by Stampfl's model. Failure to meet the requirements of the model might account for the failures to replicate. The present study included physiological and subjective measures of fear during therapy in an attempt to replicate the Kirchner and Hogan (1966) experiment, using double-blind methodology and a placebo control. It was expected that after exposure to a tape recorded session of implosive therapy, rat phobic subjects would demonstrate significantly greater improvement than controls on a behavioral avoidance test and a questionnaire. In conformity with the extinction model, it was predicted that the implosion group would manifest greater average fear, as well as a significant reduction in fear, during therapy, and that measures of fear during therapy and at assessment would he correlated.