Background: Matrix metalloproteinases (MMP) in blood are promising new diagnostic tools. It was shown that the blood sampling process resulted in different blood concentrations of MMPs. To clarify whether the sampling process also influences the diagnostic validity of MMPs, MMP-9 measurements were performed in plasma and serum samples of patients with prostate carcinoma and renal cell cancer. Methods: MMP-9 ELISAs were performed in samples of heparin plasma and serum collected in blood tubes with and without clot accelerator. Measurements were undertaken in 78 healthy persons, 33 patients with prostate carcinoma and 33 patients with renal cell carcinoma. Results: MMP-9 showed higher concentrations in serum samples than in heparin plasma and was about threefold higher in serum samples collected in tubes with clot activator than in native serum samples. Both patient groups had lower MMP-9 concentrations in serum, whereas in plasma, patients with renal cell carcinoma had higher, but patients with prostate cancer unchanged MMP-9 concentrations. 13 of 33 patients with renal cell carcinoma had increased MMP-9 plasma values but no patient had increased serum concentrations. Conclusions: To optimise the diagnostic validity of the MMP-9 in blood, measurements should be performed in heparin plasma but not in serum. (C) 2001 Elsevier Science B.V. All rights reserved.