Human semaphorin 4D (SEMA4D), a type I integral membrane glycoprotein, regulates key cellular functions (e.g. cell-cell communication, platelet activation). Its 120 kDa extracellular region can be shed from the membrane to release soluble SEMA4D (sSEMA4D). Studies on circulating sSEMA4D levels are mostly performed with poorly characterized assays and use serum and plasma as matrix. We developed and validated a sandwich ELISA utilizing two monoclonal antibodies with resolved epitopes and determined affinities. Human serum and plasma samples were analyzed, and the influence of protease activity on sSEMA4D concentration was tested by collecting samples in the presence of the protease inhibitor TAPI-1. Both antibodies recognize conformational epitopes in the sema domain. Validation for plasma (EDTA, citrate, heparin) showed valid specificity, precision, accuracy, dilution linearity, and robustness. The assay shows a calibration range from 62.5 to 2000 pmol/L with a quantification limit of 31 pmol/L. sSEMA4D was significantly higher in serum than in plasma, whereas serum and plasma levels from samples collected in the presence of TAPI-1 showed no statistical difference. This ELISA provides a reliable tool for the quantification of sSEMA4D in human plasma. Serum is not recommended as matrix due to the accumulation of shed SEMA4D during blood coagulation altering serum sSEMA4D levels.