Cancer invasion and metastasis depend on tumor-induced angiogenesis, which is the means by which cancer cells attract and maintain a blood supply. The cellular processes that occur during angiogenesis are tightly coordinated by signaling molecules. Understanding how blood vessel cells synthesize multiple biochemical signals can lead to the development of novel therapeutic strategies to combat cancer. This study is the first to propose a signal transduction model highlighting the cross-talk between key receptors involved in angiogenesis, namely growth factor, integrin, and cadherin receptors. From experimental data, we construct a stochastic Boolean network model of receptor cross-talk, and analyze the dynamical stability of the network. We derive a robust mapping that links external stimuli to cell phenotypes and show that cross-talk is crucial to phenotype determination. This mapping is extraordinarily stable against molecular noise with one important exception: an oscillatory feedback loop between the signaling molecules RhoA and Rac1 is unstable under arbitrarily low noise, and leads to erratic, dysfunctional cell motion. The model yields a surprising result of a "go-and-grow" phenotype, contradicting the popular "go-or-grow" hypothesis.