Analyzing surface epitopes of single HIV particles holds great potential for the development of vaccine candidates. However, existing technologies do not allow corresponding screens at high throughput. We present here a single-virus droplet-based microfluidics platform enabling sorting of millions of HIV-1 particles with >99% efficiency, based on the expression of epitopes recognized by broadly neutralizing antibodies. We show that virus particles displaying these epitopes can be identified, sorted, and analyzed by next-generation sequencing: an approximately 1,900-fold enrichment of viral particles displaying neutralizing epitopes could be obtained in a single sort, thus opening the way for screening diverse virus libraries with optimal antigenic features for HIV vaccine candidates.