Critical challenges remain in clinical translation of extracellular vesicle (EV)-based therapeutics due to the absence of methods to enrich cells with high EV secretion. Current cell sorting methods are limited to surface markers that are uncorrelated to EV secretion or therapeutic potential. Here, we utilize a nanovial technology for enrichment of millions of single cells based on EV secretion. This approach is applied to select mesenchymal stem cells (MSCs) with high EV secretion as therapeutic cells for improving treatment. The selected MSCs exhibit distinct transcriptional profiles associated with EV biogenesis and vascular regeneration and maintain high levels of EV secretion after sorting and regrowth. In a mouse model of myocardial infarction, treatment with high-secreting MSCs improves heart functions compared to treatment with low-secreting MSCs. These findings highlight the therapeutic importance of EV secretion in regenerative cell therapies and suggest that selecting cells based on EV secretion could enhance therapeutic efficacy.