Proteins, although structurally unsuitable for their own replication, are the structures that translate an abstract DNA sequence into real action, making possible catalysis, signaling and supramolecular morphologies. Protein targets have most often been recognized by starting from antineoplastic effects initially observed at the level of an organism or a cell and working in a reductionist manner. However, no oncoproteins have been identified as cancer chemotherapy targets through this downward approach; it has yielded only targets at low levels within the replication machinery, never at the level of proliferation control and signaling proteins. Drug-like molecules with molecular weights around 500 seem to work well as inhibitors of thyrosine kinases, docking in their ATP pocket; however, molecules of this kind are perhaps too compact for selective interference at the level of cross-talks between the domains of two signaling proteins. Rapid evolution has recently taken place in possibilities for internalizing amphipathic molecules of molecular weights of several thousand. Under these conditions it becomes possible to conceive peptidomimetic drugs of several dozen amino-acid-like units, and these are much better suited for such selective interference. Small, dedicated peptidomimetic combinatorial libraries, based on a natural protein motif involved in the cross-talk of interest and on computer modeling, are natural allies in this game. A recently developed peptidomimetic lead capable of interfering with Myc activity1,2 will be discussed within the framework of this new strategy for developing antineoplastic drugs.