Because invasion is, either directly or via metastasis formation, the main cause of death in cancer patients, development of efficient anti-invasive agents is an important research challenge. We have established a screening program for potentially anti-invasive compounds. The assay is based on organotypic confronting cultures between human invasive cancer cells and a fragment of normal tissue in three dimensions. Anti-invasive agents appeared to be heterogeneous with regard to their chemical nature, but plant alkaloids, polyphenolics and some of their synthetic congeners were well represented. Even within this group, active compounds were quite diverse: (+)-catechin, tangeretin, xanthohumol and other prenylated chalcones, 3,7-dimethoxyflavone, a pyrazole derivative, an isoxazolylcoumarin and a prenylated desoxybenzoin. The data gathered in this system are now applied in two projects. Firstly, structure-activity relationships are explored with computer models using an artificial neural network approach, based on quantitative structural descriptors. The aim of this study is the prediction and design of optimally efficient anti-invasive compounds. Secondly, the metabolism of orally ingested plant polyphenolics by colonic bacteria is studied in a simulator of the human intestinal microbial ecosystem (SHIME) and in human intervention trials. This method should provide information on the final bioavailability of the active compounds in the human body, with regard to microbial metabolism, and the feasibility of designing pre- or probiotics that increase the generation of active principles for absorption in the gastro-intestinal tract. The final and global aim of all these studies is to predict, synthesize and apply in vivo molecules with an optimal anti-invasive, and hence an anti-metastatic activity against cancer.