Tumour-derived exosomes can be released to serum and provide information on the features of the malignancy, however, in order to perform systematic studies in biological samples, faster diagnostic techniques are needed, especially for detection of low abundance proteins. Most human cancer cells are positive for at least one ligand for the activating immune receptor NKG2D and the presence in plasma of NKG2D-ligands can be associated with prognosis.Using MICA as example of a tumour-derived antigen, endogenously expressed in metastatic melanoma and recruited to exosomes, we have developed two immunocapture-based assays for detection of different epitopes in nanovesicles. Although both techniques, enzyme-linked immunosorbent assay (ELISA) and Lateral flow immunoassays (LFIA) have the same theoretical basis, that is, using capture and detection antibodies for a colorimetric read-out, analysis of exosome-bound proteins poses methodological problems that do not occur when these techniques are used for detection of soluble molecules, due to the presence of multiple epitopes on the vesicle.Here we demonstrate that, in ELISA, the signal obtained was directly proportional to the amount of epitopes per exosome. In LFIA, the amount of detection antibody immobilized in Au-nanoparticles needs to be low for efficient detection, otherwise steric hindrance results in lower signal. We describe the conditions for detection of MICA in exosomes and prove, for the first time using both techniques, the co-existence in one vesicle of exosomal markers (the tetraspanins CD9, CD63 and CD81) and an endogenously expressed tumour-derived antigen. The study also reveals that scarce proteins can be used as targets for detection antibody in LFIA with a better result than very abundant proteins and that the conditions can be optimized for detection of the protein in plasma.These results open the possibility of analyzing biological samples for the presence of tumour-derived exosomes using high throughput techniques.