Current approaches for the fluorescent labelling of extracellular vesicles (EVs) have been reported to produce widely variable and controversial results, highlighting a significant need for validated, reproducible labelling methods to advance the field of EV research. Lipophilic membrane dyes are commonly used but have been shown to produce non-specific fluorescent particles that are indistinguishable from labelled EVs, confounding experimental results. We aimed to distinguish conditions that can either promote or reduce the formation of non-specific dye particles when using the prototypical lipophilic membrane dye PKH26. We optimised a labelling approach that minimises the production of non-specific dye particles by altering buffer conditions during staining and validated this method across cell-based and in vivo systems of EV biodistribution. To do this, we specifically isolated small EVs using ultrafiltration and size exclusion chromatography and validated sample purity and post-isolation processing steps. We then used single-EV spectral flow cytometry and transmission electron microscopy to investigate the impact of four different buffer conditions on PKH26 non-specific particle formation. We also determined the extent to which non-specific PKH26 particles were detectable in cell-based assays and in vivo within mouse lymph nodes using flow cytometry, immunofluorescence, and intravital imaging. By optimising buffer conditions to eliminate additional protein, we were able to minimise the formation of dye aggregates while maintaining efficient EV labelling, producing a much higher signal-to-noise ratio both in vitro and in vivo. We also demonstrate that failure to include proper vehicle controls can have significant implications on experimental results, leading to false positive data. This work emphasizes the importance of adequately benchmarking EV labelling approaches as it is essential for accurate evaluation of EV trafficking in physiologic and pathologic states.