Proteins identified in biological fluids of cancer patients could be helpful for both diagnosis and prognosis. However, clinical proteomics based on analysis of protein profiles in biological fluids has demonstrated various flaws, most of them related to the difficulties met in reproducibility. These difficulties could be partly overcome by accurate standardisation of pre-analytical and analytical steps of these studies. The size of the patient cohort is one of the parameters that determine the powerfulness of the study. Recruitment of a cohort with a sufficient size often implies multicentric studies in which analysis of the reproducibility between centres and standardisation of pre-analytical and analytical steps are essential. Such a standardisation requires the use of calibrated samples as common references.