The central benzodiazepine receptor (cBZr) has long been implicated in anxiety disorders on the basis of: (i) the well-known anxiolytic and anxiogenic properties of cBZr agonists and inverse agonists, respectively; (ii) a possibly reduced sensitivity to benzodiazepines in anxious subjects; and (iii) a putative endogenous ligand. Thus, two main hypothesis have been advanced, namely changes in the concentration or properties of the latter, and changes in the GABAA complex conformation, which contains the cBZr. Neither postmortem studies nor appropriate animal models are available to investigate these ideas. We have used positron emission tomography (PET) to measure both the density and affinity of the cBZr in multiple brain regions in unmedicated patients and age- and sex-matched healthy volunteers, and have looked for differences between groups as well as correlations between cBZr parameters and state and trait anxiety scores. We studied 10 unmedicated patients (sex ratio 1 : 1; mean age: 39 years), prospectively recruited using DSM III-R criteria, and 10 age- and gender-matched healthy unmedicated volunteers. Thanks to a PET procedure using two successive administrations of 11C-flumazenil (at high and low specific radioactivity) and previously validated by us, we estimated the Bmax, Kd and bound : free (B/F) ratios in 11 neocortical areas and in the cerebellum. Before and after the PET session, anxiety scores from Spielberger's and Covi's scales were obtained. There was no statistically significant difference in Bmax, Kd or B/F-values between the two groups for any region. Across the two groups, there were only a few marginally significant anxiety-score-PET correlations, suggesting chance findings. This is the first fully quantitative study to report on the relationships between cBZr parameters and anxiety. Using two independent approaches (i.e. group comparison and across-group correlations), we found no evidence for a link between anxiety trait or state and the cBZr in neocortex or cerebellum in this sample. These findings, if confirmed by studies on larger samples, have implications for the pharmacotherapy of anxiety disorders, and will need to be considered when designing new neurobiological models of anxiety.