Background: Degradation of the inhibitory neurotransmitter γ-aminobutyric acid (GABA) is mainly catalysed by GABA aminotransferase (GABA-AT), excessive activity of which leads to convulsions. Inhibition of GABA-AT increases the concentration of GABA and can terminate the convulsions. Several studies have revealed that GABA analogues could be the outstanding scaffolds for the design of potent inhibitors of GABA-AT. The poor ability of GABA analogues to cross the blood–brain barrier (BBB), always produces low therapeutic index.` However, Vigabatrin, a mechanism-based inhibitor of GABA-AT, is currently approved treatment of epilepsy, but it has harmful side effects, leaving a need for improved GABA-AT inactivators.Experimental design: In our present in silico investigation, AutoDock 4.2,-based on Lamarckian genetic algorithm was employed for virtual screen of a compound library with 35 entries (Schiff’s bases of GABA) in search for novel and selective inhibitors of GABA-AT. Results: By means of flexible type of molecular docking, we proposed that these designed molecules could successfully bind into the active pocket of GABA-AT with good predicted affinities in comparison to standard vigabatrin. Among the designed analogues, HIG18, HIG28 and HIG30 showed significant binding free energy of -10.25, -9.88 and -9.31 kcal/mol with predicted inhibitory constant values of 0.03, 0.05 and 0.15 μM respectively. Conclusion: Using ligand-based drug design, we proposed that electron withdrawing phenyl substituted heterocyclic imines of GABA could be considered as promising structures for synthesis and testing of new GABA-AT inhibitors from this class. We hypothesize that novel GABA analogues with an azomethine linkage incorporated with heterocyclic system can have increased affinity and more lipophilic character that would provide a probability of having less toxic effect in the therapy of convulsions.