Extracellular vesicles (EVs) have attracted significant attention as impactful diagnostic biomarkers, since their properties are closely related to specific clinical conditions. However, designing experiments that involve EVs phenotyping is usually highly challenging and time-consuming, due to laborious optimization steps that require very long or even overnight incubation durations. In this work, we demonstrate label-free, real-time detection and phenotyping of extracellular vesicles binding to a multiplexed surface. With the ability of label-free kinetic binding measurements using the Interferometric Reflectance Imaging Sensor (IRIS) in a microfluidic chamber, we successfully optimize the capture reaction by tuning various assay conditions (incubation time, flow conditions, surface probe density and specificity). A single (less than 1 hour) experiment allows for characterization of binding affinities of the EVs to multiplexed probes. We demonstrate kinetic characterization of 18 different probe conditions, namely three different antobodies, each spotted at six different concentrations, simultaneously. The affinity characterization is then analyzed through a model which considers the complexity of multivalent binding of large structures to a carpet of probes, and therefore introduces a combination of fast and slow association and dissociation parameters. Additionally, our results confirm higher affinity of EVs to aCD81 with respect to aCD9 and aCD63. Single-vesicle imaging measurements corroborate our findings, as well as confirming the EVs nature of the captured particles through fluorescence staining of the EVs membrane and cargo.