Recently, accumulated statistical data indicate the protective effect of caffeine consumption against several types of cancer diseases. There are also reports about protective effect of caffeine and other xanthines against tumors induced by polycyclic aromatic hydrocarbons. One of the explanations is based on biological activation of such carcinogens by cytochromes that are also known for metabolism of caffeine. However, there is also numerous data indicating reverse effect on cytotoxicity of anticancer drugs that inhibit the action of topoisomerase I (e.g. Camptothecin or Topotecan) and topoisomerase II inhibitors (e.g. Doxorubicin, Mitoxantrone or mAMSA). In this work we tested the hypothesis that the caffeine protective effect is the result of sequestering of aromatic mutagens by formation of stacking (pi-pi) complexes. As the models for the study we have chosen two well-known mutagens, that do not require metabolical activation: quinacrine mustard(QM, aromatic, heterocyclic nitrogen mustard) and mechlorethamine (NM2, aliphatic nitrogen mustard). The flow cytometry study of these agents' action on the cell cycle of HL-60 cells indicated that caffeine prevents the cytotoxic action of QM, but not that of NM2. The formations of stacking complexes of QM with caffeine were confirmed by light absorption, calorimetric measurements and by molecular modeling calculation. Using the statistical thermodynamics calculations we calculated the "neighborhood" association constant (K(AC)=59+/-2M(-1)) and enthalpy change (DeltaH(0')=-116cal mol(-1)); the favorable entropy change of complex formation (DeltaS(0')=7.72cal mol(-1)K(-1), due to release of several water molecules, associated with components in the process of complex formation). The Gibbs' free energy change of QM-CAF formation is DeltaG(0')=-2.41kcal mol(-1). We were unable to detect any interaction between NM2 and caffeine either by spectroscopic or calorimetric measurement. In order to establish, whether the intercalation of QM plays any role in cytotoxic effect we tested, as a control, non-alkylatiatig, but also intercalating QM derivative-quinacrine (Q). The later had no cytostatic effect on HL-60 cell even at there order of higher concentration than QM or NM2 but, similar to QM forms (which we demonstrated) stacking complexes with caffeine (K(AC)=75+/-3M(-1)). These results strongly indicate, that the attenuating effect of caffeine on cytotoxic or mutagenic effects of some mutagens, is not the results of metabolic processes in the cells, but simply the physicochemical process of sequestering of aromatic molecules (potential carcinogens or mutagens) by formation of stacking complexes with them. The caffeine may then act as the "interceptor" of potential carcinogens (especially in the upper part of digesting track where its concentration can reach the concentration of mM level). There is, however, no indication either in the literature or in our experiments that xanthines can reverse the damage to nucleic acids when the damage to DNA has already occurred.