The substances inhibiting cytochrome Р450 1А2 (CYP1A2) represent a perspective class of new drugs, which application in clinical practice can become the important part in preventive maintenance in oncology. The present work is devoted to computer modelling of 3-D structure of CYP1A2 and searching of new inhibitors by database mining. The modelling of CYP1A2 was done based on homology with 4 bacterial cytochromes P450 with known 3-D structure. For optimization of CYP1A2 active site structure the models of its complexes with characteristic substrates (caffeine and 7-ethoxyresorufin) were designed. These complexes were optimized by molecular dynamics simulation in water. The models of 24 complexes of CYP1A2 with known ligands with known Kd were designed by means of DockSearch and LeapFrog programs. 3D-QSAR model with good predictive force was created based on these complexes. On a final stage the search of knew CYP1A2 ligands in testing database (more than 23.000 substances from database Maybridge and 112 known CYP1A2 ligands from database Metabolite, MDL) was executed. 680 potential ligands of CYP1A2 with Kd values, comparable with known ones were obtained. This number has included 73 compounds from 112 known ligands, introduced in tested database as the internal control.