AbstractSmall extracellular vesicles (sEV) purified from blood have great potential clinically as biomarkers for systemic disease; however interpretation is complicated by release of sEV ex vivo after blood taking. To quantify the problem and devise ways to minimise it, we characterised sEV in paired serum, plasma and platelet poor plasma (PPP) samples from healthy donors. Immunoblotting showed twofold greater abundance of CD9 in sEV fractions from fresh serum than from fresh plasma or PPP. MACSPlex confirmed this, and showed that proteins expressed on platelet sEV, either exclusively (CD41b, CD42a and CD62P) or more widely (HLA‐ABC, CD24, CD29 and CD31) were also twofold more abundant; by contrast non‐platelet proteins (including CD81) were no different. Storage of plasma (but not serum) increased abundance of platelet and selected leukocyte sEV proteins to at least that of serum, and this could be recapitulated by activating cells in fresh plasma by Ca2+, an effect abrogated in PPP. This suggests that a substantial proportion of sEV in serum and stored plasma were generated ex vivo, which is not the case for fresh plasma or PPP. Thus we provide strategies to minimise ex vivo sEV generation and criteria for identifying those that were present in vivo.