The comparison of pharmacokinetics of DNR in mouse plasma, in the DNR naturally resistant B16 melanoma and in the DNR naturally sensitive P388 leukemia showed that there is no direct correlation between total concentrations of this drug in tumours and the sensitivity resistance of these tissues. A finding which demonstrates the inadequacy of distribution models to select new potential anticancer drugs. Cytotoxicity of DNR and its metabolites to B16 melanoma and P388 leukemia cell lines were determined in vitro. Calculated inhibitory concentrations 50 (IC50) were compared to maximal concentrations determined by pharmacokinetic studies. In all cases in vitro IC50 were lower than Cmax values. Moreover, resistant cells in vivo were found to be sensitive to DNR and metabolites when they are propagated in vitro. Tissue concentrations, as well as in vitro data, were fitted to appropriate models by an original program (FADHA) which uses the simplex method to minimize a non-linear cost function. Best fit models were chosen by statistical criteria.