A proportion of patients with panic disorder (PD) display an increased sensitivity to the anxiogenic/panicogenic properties of caffeine. The aim of this study is to identify probable baseline differences between PD patients who panic and those who do not, after caffeine administration. In a randomized, double-blind, cross-over experiment performed in two occasions 3-7 days apart, 200 and 400 mg of caffeine, respectively, were administered in a coffee form to 23 patients with PD with or without Agoraphobia. Evaluations included the State-Trait Anxiety Inventory, the DSM-IV 'panic attack' symptoms (visual analogue scale form), the Symptom Checklist-90-Revised (SCL-90-R), as well as breath-holding (BH) duration, heartbeat perception accuracy and heart rate. Only those patients who did not present a panic attack after both challenges ('no panic group', N=14, 66.7%), and those who presented a panic attack after at least one challenge ('panic group', n=7, 33.3%) were included in the analysis. The panickers, compared to the non-panickers, presented at baseline: significantly higher total score of the SCL-90-R; significantly higher scores on all the SCL-90-R clusters of symptoms, except that of 'paranoid ideation'; significantly lower BH duration. The present preliminary findings indicate that PD patients who panic after a 200 mg or a 400 mg caffeine challenge, compared to the PD patients who do not panic after both of these challenges, may present at baseline significantly higher non-specific general psychopathology--as reflected in the SCL-90-R--and significantly shorter BH duration.