Inter-receptor interactions play a key role in receptor signaling, which is the first step in cell signaling in response to external stimuli. In the case of Vascular Endothelial Growth Factor Receptor 2 (VEGFR2), dimerization is necessary for activation. VEGFR2 undergoes reversible interactions also in the absence of ligand. For a quantitative understanding of transmembrane signal transduction, it is necessary to quantify the interaction kinetics of VEGFR2 on the cell surface. Live-cell single-molecule imaging (SMI) has the powerful ability to capture receptor interaction events in their native cellular environment with high spatiotemporal resolution. However, it reveals these interactions for only the labeled subset, which is a small fraction of the full population of receptors. We have previously shown that mathematical modeling, combined with SMI data, offers a route to compensate for this lost information and infer the population-level receptor interaction kinetics from SMI data. Here, we applied this approach to VEGFR2, both wildtype full length VEGFR2 (FLR2) and a truncated mutant consisting of its extracellular and transmembrane domains (ECTM), which has been shown to exhibit reduced homotypic interactions in the absence of ligand. We developed a stochastic mathematical model mimicking VEGFR2 diffusion and interactions and determined the unknown model parameters through a combination of direct experimental measurements and stochastic model calibration. We found that a model of dimerization was sufficient to describe VEGFR2 interactions in the absence of ligand. While ECTM was primarily monomeric, FLR2 exhibited a substantial fraction of dimers. Our inference revealed that the difference between FLR2 and ECTM was primarily in the dimer association rate constant, which was about an order of magnitude lower for ECTM than for FLR2. To our knowledge, this is the first time that the interaction kinetics of VEGFR2 have been calculated in live cells.