The biochemical processes occurring within a living cell span a spectrum of scales in space and time, ranging from the nano- to the macro-scale. We note that a single cellular process often operates on multiple spatial and temporal scales, and thus it becomes necessary to combine modeling techniques in multiscale approaches, in which different levels of theory are synergized to describe a system at a number of scales or resolutions. In this work we apply a multiscale modeling framework to investigate the molecular regulatory mechanisms governing the activation of the ErbB receptor tyrosine kinases, a family of kinases which are commonly over-expressed or mutated in human cancers, with a focus on the HER3 and HER4 kinases. Our multiscale model of HER3, a kinase which, until recently, has been considered kinase-dead, presents evidence of HER3 catalytic activity and demonstrates that even a weak HER3 signal can be amplified by other cellular signaling mechanisms to induce drug resistance to tyrosine kinase inhibitors in silico. Thus HER3, rather than the commonly-targeted EGFR and HER2 kinases, may represent a superior therapeutic target in specific ErbB-driven cancers. In the second major study, we construct a multiscale model of activity in the HER4 kinase, which has been shown to perform an anti-cancer role in certain tumor cells, by steering the cell toward a program of cellular differentiation and away from a program of uncontrolled proliferation. Our HER4 model, which applies a combined computational and experimental approach, elucidates the molecular mechanisms underlying this HER4-mediated ‘switch’ to the cellular differentiation program, with the ultimate aim of exploiting or modulating the HER4 pathway as a potential therapy in specific ErbB-driven cancers. Furthermore, the model provides structural insights into the effects of several HER4 somatic mutations which have recently been discovered in a subset of cancer patients, and which abrogate the anti-cancer effects of HER4 activity. We have illustrated that multiscale modeling provides a powerful and quantitative platform for investigating the complexity inherent in intracellular signaling pathways and rationalizing the effects of molecular perturbations on downstream signaling events and ultimately, on the cell phenotype.