A combined graph describing the growth inhibition values from a number of human cancer cell lines represents an activity profile for a compound. The fact that compounds with similar activity profiles often show similar Mode of Action (MOA) has frequently been used in prediction of MOA. Obtaining the profiles is demanding with respect to both time and resources. Therefore, as a work and time efficient alternative, we explore the central premise of medicinal chemistry that structurally similar molecules often have similar biological activity. In this study we investigate correlations between chemical structure and MOA, and subsequently use this as a complementing basis for prediction. The correlations between MOA and activity profile on one hand and between MOA and chemical structure on the other were analyzed for anticancer agents, classified with regard to MOA, using Principal Component Analysis (PCA), chemographic mapping with ChemGPS-NP, and Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA). The compounds clustered according to MOA both based on chemical structures and activity profiles. The subsequent validation with external test sets showed that initial PCA scores prediction or chemographic mapping followed by OPLS-DA could be used for the prediction of MOA as well as identification of novel MOAs in a highly accurate way. An efficient and straight forward procedure for the prediction of MOA of anticancer agents is suggested. With todays resistance problems in cancer therapy, there is a need for new anticancer agents and mechanisms. We believe that the fast initial virtual guidance this procedure implies, especially the novel step using ChemGPS-NP, could be of general use in early stages of cancer research.