We compared two protocols for the expansion of human mesenchymal stromal cells (hMSCs) starting from diagnostic samples of BM aspirates (2-5 ml) or using the remnants in the bag and filter at the end of the BM infusions. The protocols differed in the presence of either 10% fetal bovine serum (FBS) or 5% platelet lysate (PL). We obtained a significantly (P=0.02) better expansion with PL, obtaining a median 1010-fold compared to 198-fold with a selected batch of FBS and in fewer days (29.8 in PL versus 41.4 in FBS). Overall, we recovered a variable number from 54.8 x 10(6) to 365 x 10(6) hMSCs in PL versus a variable number from 2.7 x 10(6) to 31 x 10(6) in FBS. No difference could be found in terms of gross morphology, differentiation potential, surface markers and immunological properties (inhibition of allogeneic PHA response and mixed lymphocyte reaction) of cells expanded with PL or FBS. The preparations were found within the range of acceptability for all the quality control criteria. Due to the clinical grade nature of the PL and the reproducibility of separate preparations, we propose this method to obtain hMSCs even from minute amounts of BM cells.