Cells have evolved biomolecular networks that process and respond to changing chemical environments. Understanding how complex protein interactions give rise to emergent network properties requires time-resolved analysis of cellular response under a large number of genetic perturbations and chemical environments. To date, the lack of technologies for scalable cell analysis under well-controlled and time-varying conditions has made such global studies either impossible or impractical. To address this need, we have developed a high-throughput microfluidic imaging platform for single-cell studies of network response under hundreds of combined genetic perturbations and time-varying stimulant sequences. Our platform combines programmable on-chip mixing and perfusion with high-throughput image acquisition and processing to perform 256 simultaneous time-lapse live-cell imaging experiments. Nonadherent cells are captured in an array of 2,048 microfluidic cell traps to allow for the imaging of eight different genotypes over 12 h and in response to 32 unique sequences of stimulation, generating a total of 49,000 images per run. Using 12 devices, we carried out >3,000 live-cell imaging experiments to investigate the mating pheromone response in Saccharomyces cerevisiae under combined genetic perturbations and changing environmental conditions. Comprehensive analysis of 11 deletion mutants reveals both distinct thresholds for morphological switching and new dynamic phenotypes that are not observed in static conditions. For example, kss1Delta, fus3Delta, msg5Delta, and ptp2Delta mutants exhibit distinctive stimulus-frequency-dependent signaling phenotypes, implicating their role in filtering and network memory. The combination of parallel microfluidic control with high-throughput imaging provides a powerful tool for systems-level studies of single-cell decision making.