The interaction of 20 nonsteroidal anti-inflammatory drugs with a model protein was studied by reversed-phase high-performance liquid chromatography and the relationship between the strength of interaction and the physicochemical parameters of drugs was elucidated by principal component analysis followed by modified nonlinear mapping and cluster analysis. It was established that the polarity and and sterical parameters of drugs exert the highest influence on the interaction. Cluster analysis and nonlinear mapping using the absolute values of principal component loadings explain more precisely the interaction than the traditional nonlinear mapping and cluster analysis do.