A cognitive explanation of the association between acute hyperventilation and panic attacks has been proposed: the extent to which sensations produced by hyperventilation are interpreted in a negative and catastrophic way is said to be a major determinant of panic. Non-clinical subjects were provided with a negative or a positive interpretation of the sensations produced by equivalent amounts of voluntary hyperventilation. As predicted, there was a significant difference between positive and negative interpretation conditions on ratings of positive and negative affect. Subjects in the positive interpretation condition experienced hyperventilation as pleasant, and subjects in the negative interpretation condition experienced hyperventilation as unpleasant, even though both groups experienced similar bodily sensations and did not differ in their prior expectations of the affective consequences of hyperventilation. When the subjects were given a positive interpretation, the number of their sensations correlated with positive affect; when a negative interpretation was given, the number of bodily sensations correlated with negative affect. The results provide support for a cognitive model of panic and are inconsistent with the view that panic is simply a symptom of hyperventilation syndrome.