The benzodiazepines (BZ) bind to the GABA A receptor (GABA A R) at the interface of its α-/γ-subunits and exert pharmacological activity as allosteric modulators. However, the distribution of the six distinct α-subunits (α 1 –α 6) in the human brain has not been mapped in detail, primarily due to lack of α-subunit selective radioligands. AZD7325 and AZD6280 were two drug candidates with partial α-subunit selectivity in vitro , in development for the treatment of anxiety. GABA A R occupancy of both drugs was examined in the human brain using [ 11 C]flumazenil PET imaging, which visualizes GABA A Rs containing the α 1 -, α 2 -, α 3 -, or α 5 -subunits with similar sensitivity. Importantly, the pattern of occupancy was heterogeneous across brain regions and different between the two drugs. This observation encouraged us to extend the analysis in an attempt to generate tentative maps of α-subunits in the human brain. Parametric images of [ 11 C]flumazenil binding in 12 subjects, obtained at baseline and following administration of different doses of AZD7325 or AZD6280, were entered into a comprehensive analysis to identify GABA A R occupancy components of the two drugs. The major outcome parameters of the fitted models were maps of the contributions of these components to the overall occupancy and binding. The maps were then explored in terms of gross anatomy and were correlated with gene expression data for the relevant α-subunits to speculate on possible α-subunit identity of the derived components. The overall occupancy was disentangled into three distinct components (C 1 to C 3) by the preferred model. C 1 was occupied by both drugs, C 2 was only occupied by AZD7325, and C 3 was not occupied by either drug. The patterns of component-specific contributions were diverse and complex, dissimilar to each other and to the overall [ 11 C]flumazenil binding. Of the three components, C 1 had the highest contribution throughout most of the brain except some cerebral nuclei, such as amygdala. The contribution of C 2 was notable in cortex and basal ganglia, and very low in thalamus and brain stem. Within the cortex, the contribution of C 3 was localized with highest values in sharply demarcated areas of the limbic, cingulate, and insular cortex. Otherwise, it had the highest contribution among components in some subcortical nuclei, was behind C 1 in thalamus, and was negligible in brain stem. All three components had a high-degree, statistically significant positive correlation with GABA A R α-subunit gene (GABRA) expression: C 1 foremost with GABRA1, C 2 foremost with GABRA2, and C 3 foremost with GABRA5. The correlations suggest that C 1 might correspond to the distribution of α 1 - (and possibly α 3 -), C 2 to that of α 2 -, and C 3 to that of α 5 -subunit-containing GABA A Rs, respectively. The components identified by the present analysis of occupancy patterns at the [ 11 C]flumazenil binding site provided putative in vivo maps of α-subunit-specific GABA A R distribution in the human brain. The findings demonstrate the feasibility of developing small molecules having preference for certain α-subunits, even if full selectivity was not yet achieved. Accordingly, the results should encourage and support the development of optimized, fully selective compounds to the benefit of basic research and drug development for the treatment of neurological and psychiatric conditions.