In this study we describe a new comparative molecular field analysis (CoMFA) model of dihydroquinazolinone and tetrasubstituted imidazole compounds with p38 MAPK inhibitory activity. A series of 51 (a training set of 40 and a test set of 11) dihydroquinazolinone [Bioorg. Med. Chem. Lett. 2003, 13, 277.] and tetrasubstituted imidazole [J. Med. Chem. 1999, 42, 2180.] derivatives known as p38 mitogen-activated protein kinase (p38 MAPK) selective inhibitors was studied by quantitative structure-activity relationship (3D-QSAR) analysis using comparative molecular field analysis. The 3D-QSAR models were generated and evaluated by a scheme that combines a genetic algorithm (GA) optimization with partial least squares (PLS) regression and by crossvalidation using the leave-one-out technique. The model was able to efficiently predict the activities of the compounds of the test set, suggesting that it can be used for the planning of new p38 MAPK inhibitor candidates useful to treat chronic inflammatory states.