2D and 3D QSAR studies were applied on a set of 28 diarylpyrimidine derivatives to model and understand their HIV-1 reverse transcriptase (RT) inhibitory activities. Special cares were taken to build our set of molecules according to their bioactive conformations which is crucial to elaborate good QSAR models. 2D QSAR was performed using the heuristic method in CODESSA which had led to a linear model (R (2) = 0.928 and s (2) = 0.015) between the inhibitory activity and five descriptors. CoMFA and CoMSIA models were established using SYBYL package of programs. The better predictive ability of the CoMSIA model (q (2) = 0.730) over the CoMFA model (q (2) = 0.597) was assigned to the large contribution of hydrogen-bonding interactions to the inhibitory activity. CoMSIA physicochemical properties are in agreement with the 2D QSAR descriptors. The CoMSIA PLS contour surfaces were mapped to the binding pocket of the RT and showed that the results obtained by the 2D and 3D models are in respect with the protein environment. This link permitted us to validate our model and give important insights for the structure activity interpretations. These results will guide further structural modification and prediction of new HIV-1 RT inhibitors.