The components of the plasminogen activation system have been reported to have prognostic impact in several cancer types, e.g. breast-, colon-, gastric- and lung cancer. Most of these studies have used quantification by enzyme-linked immunosorbent assay (ELISA) on tumour tissue extracts. However, results in non-small cell lung cancer (NSCLC) studies obtained by quantitative ELISA and semiquantitative immunohistochemistry differ. If the prognostic value of the components of the plasminogen activation system is to be exploited clinically in the future, it is important to choose an easy and valid methodology. In the present study we investigated levels of plasminogen activator inhibitor type 1 (PAI-1) and urokinase plasminogen activator receptor (uPAR), as quantitated by ELISA in tumour extracts from 64 NSCLC patients (38 squamous cell carcinomas, 26 adenocarcinomas), and compared them to staining intensity as semiquantitated by immunohistochemistry for PAI-1 and uPAR on corresponding cryostat sections. A significant association (r = 0.49, P < 0.0001) was found between the PAI-1 levels measured by ELISA and semiquantitated by immunohistochemistry. No association was found for uPAR. When correlating levels of PAI-1 and uPAR determined by ELISA and immunohistochemistry, respectively, to survival status, no significant correlation was found for any of the subgroups. At present neither of the methods examined in the present study can be recommended as superior for quantitating PAI-1 and uPAR with the aim of predicting prognosis. In conclusion, a larger comparative study is needed to clarify the relationship between ELISA and immunohistochemical results, before a methodology for clinical use can be chosen in non-small cell lung cancer.