Extracellular vesicles (EVs) are nanostructures that are used as sources of biomarkers. To better understand whether EVs could be exploited as diagnostic and prognostic biomarkers in Amyotrophic Lateral Sclerosis (ALS), we analyzed plasma-derived of ALS patients and relative healthy and diseased controls. Using the nickel-based isolation, a recently published EV purification method, we unmasked peculiar features in plasma EVs of ALS patients with a potential straightforward application in a clinical setting. We report that the number of particles is increased in the plasma of ALS patients and of two mouse models of ALS while the average diameter is decreased. Proteins like HSP90 and phosphorylated TDP-43 are differentially represented in ALS patients and mice compared to the controls. In terms of disease progression, the levels of phosphorylated TDP-43 and cyclophilin A, along with the EV size distribution discriminated fast and slow progressors of the diseases suggesting a new means for patient stratification. We exploited the EV size distribution with machine learning analysis that combining different EV parameters resulted in very high prediction rates for disease diagnosis and prognosis