Anti-cancer drugs that target the mutationally activated drivers of malignancy have given responses in solid tumors that generally are incomplete and transient. Resistance to these therapies appears to be due in part to the fact that signaling systems operate via networks with redundant elements capable of compensating for inhibition of single targets. Thus, it is likely that effective deployment of targeted therapies will require the development of combination therapies that can co-target these compensatory pathways. We have used global analytical approaches (gene expression, reverse phase protein arrays) and combinatorial small molecule screens (37 lines, 4 lineages, 70 inhibitors) to identify pathways whose activity can compensate for inhibition of the MAP Kinase pathway and thus are targets for combination therapies. Several promising combinations have been identified, some predictable and others quite surprising. Among the predictable results was synergistic cytotoxicity from combining inhibitors of the PI3K and MAPK pathways. Inhibition of PI3K led to enhanced MAPK signaling, and the reverse. We have identified p70S6 Kinase as an important convergence point for the PI3K and MAPK pathways. Combination therapy uniquely caused synergistic inhibition of p70S6K activity when synergistic cytotoxicity was induced, and expression of mutationally activated P70S6K could rescue this effect. This identifies p70S6K as a critical node in these pathways, and a potentially important single target. Among the surprising results was the diversity of synergistic combinations in a panel of 16 B-Raf mutant melanomas. Although B-Raf mutational status predicted sensitivity to the B-Raf inhibitor vemurafenib, each line differed in the drugs that induced synergistic cytotoxicity. Thus, B-Raf is “wired into” the signaling network differently in each of these cells, presumably a consequence of the diverse array of secondary mutations. Collectively, these data point out the extraordinary robustness and plasticity of cancer cell signaling networks, and the challenges of individualizing therapies.