n certain contexts, different cellular responses arise from variations in protein expression levels or activity. Using single- cell measurements and computational models, we recently showed how variability in protein expression levels is responsible for cell-to- cell variability in the timing of apoptosis in TRAIL-treated cells. Still, predicting cellular phenotypes is challenging when multiple proteins and complex dynamics are involved. Here I will present an approach, based on Lyapunov analysis of differential equation models, which allows us to compute phase diagrams mapping predicted cellular phenotypes to different concentrations of regulatory proteins. We validated this approach experimentally, identifying key determinants for whether cells require, or not, mitochondrial outer membrane permeabilization during cell death induced by TNF superfamily ligands (Type II vs. I phenotypes).