This study investigated the role of relevant vs irrelevant fear cues in the flooding of phobic patients. Six specific phobics and 10 agoraphobics were treated in a balanced crossover design. Eight patients had eight sessions of imaginal flooding concerned with their phobias followed by eight imaginal sessions concerned with situations which are normally frightening to anybody. Another eight patients had the same two treatments in the reverse order. The combined effects of both treatments after 16 sessions resulted in significant improvement on clinical, attitudinal, and heart-rate measures. Improvement was maintained at six months follow-up. Eight sessions by each treatment alone also produced significant improvement on clinical and attitudinal measures. Irrelevant fear also produced significant improvement in heart-rate and skin-conductance measures. The two treatments did not differ significantly from each other in their effects, except that irrelevant fear produced significantly more improvement than did relevant flooding in subjective anxiety during phobic imagery. The two treatments had significantly different prognostic correlates. Heightened physiological activity at the start of treatment predicted a good outcome to relevant flooding but not to irrelevant fear. High subjective anxiety during imagery before treatment predicted poor outcome to irrelevant fear. High anxiety during treatment sessions predicted good outcome to irrelevant fear, but did not correlate with outcome to relevant flooding. The experience of relevant and irrelevant fear in fantasy reduced phobic anxiety and avoidance to a similar extent, but appeared to do so through different mechanisms. These mechanisms need not be mutually exclusive and might be additive.