Dihydrofolate reductase (DHFR) has been used as a target for antimicrobial drug discovery against a variety of pathogenic microorganisms, including opportunistic microorganisms; Pneumocystis carinii (pc), Toxoplasma gondii (tg) and Mycobacterium avium complex (ma). In this regard, several DHFR inhibitors are reported against pc and tg and ma. However, selectivity issue of these inhibitors over human DHFR often preclude their development and clinical use. In the first part of this work, various computational approaches including available crystallographic structures, binding affinity prediction, pharmacophore mapping, QSAR, homology modelling used for design of DHFR inhibitors against opportunistic microorganisms are reviewed, to understand specific interactions required for inhibition of microbial DHFR. Secondly, comprehensive molecular modelling techniques were used, to establish structure-chemical-feature-based pharmacophore models for pcDHFR, tgDHFR and mammalian DHFR. The results show that, the information encoded by ligand based approaches like pharmacophore mapping and 3D-QSAR methods are in well agreement with the information coded in the receptor structure. A combination of ligand and structure based approaches provides understanding of ligand-receptor interactions. The study indicated that the value of small alkyl moieties at position 5 of the bicyclic nitrogen containing nucleus along with a bulky group attached at the C-6 via suitable linker could optimize activity, with regard to both potency and selectivity.