Cellular signaling responses show substantial cell-to-cell heterogeneity, which is often ascribed to the inherent randomness of biochemical reactions, termed molecular noise, wherein high noise implies low signaling fidelity. Alternatively, heterogeneity could arise from differences in molecular content between cells, termed molecular phenotypic variability, which does not necessarily imply imprecise signaling. The contribution of these two processes to signaling heterogeneity is unclear. Here, we fused fibroblasts to produce binuclear syncytia to distinguish noise from phenotypic variability in the analysis of cytokine signaling. We reasoned that the responses of the two nuclei within one syncytium could approximate the signaling outcomes of two cells with the same molecular content, thereby disclosing noise contribution, whereas comparison of different syncytia should reveal contribution of phenotypic variability. We found that ~90% of the variance in the primary response (which was the abundance of phosphorylated, nuclear STAT) to stimulation with the cytokines interferon-γ and oncostatin M resulted from differences in the molecular content of individual cells. Thus, our data reveal that cytokine signaling in the system used here operates in a reproducible, high-fidelity manner.