Forty-nine DSM-IV diagnosed claustrophobics and 49 sex- and age-matched community controls, without any current or past psychiatric disorder, were asked to estimate the probability that three types if events would occur if they were in the described situations. The events were claustrophobic, generally negative, and positive in nature. The results showed that claustrophobics significantly overestimated the probability of events they specifically feared, i.e. the claustrophobic events, while there was no difference between the groups regarding generally negative events and positive events. This finding remained when the higher scores for claustrophobics on the Claustrophobia scale and the Anxiety Sensitivity Index were covaried out. The conclusion that can be drawn is that claustrophobics' probability ratings are characterized by distortions that are specifically connected to anxiety-arousing events and not negative events in general. The hypothesis is proposed that this may be explained by an exaggerated use of simplified rules-of-thumb for probability estimations that build on availability in memory, simulation, and representativity.