To further investigate the role of serotonin (5-HT) in anxiety, two tests were used in human subjects. The first was the conditioning of skin conductance response (CSCR) that associates a tone to a loud noise. The second was simulated public speaking (SPS), which is believed to represent unconditioned fear. In healthy volunteers the 5-HT(2A) receptor blocker and 5-HT reuptake inhibitor nefazodone reduced subjective anxiety and the number of spontaneous fluctuations of skin conductance during CSCR, but enhanced anxiety induced by SPS. Opposite effects had been reported with the 5-HT releasing and uptake-inhibiting agent D-fenfluramine. Panic patients behaved like controls in the CSCR. However, they had a higher level of baseline anxiety and were insensitive to SPS. This profile resembles the reported effect of the non-selective 5-HT receptor blocker metergoline in healthy volunteers. Therefore, panic patients seem to process unconditioned fear abnormally, which may be due to lack of 5-HT inhibition in brain structures commanding flight from proximal danger stimuli.