Matrix metalloproteinases (MMPs) are involved in tumor growth and spreading. Here, we investigated the tumor immunoreactive protein of MMP-2, MMP-9 and TIMP-1 as well as the levels of circulating total TIMP-1 and MMP-2/TIMP-2-complex as prognostic factors in lung cancer patients. The material included 59 patients, 30 with a squamous cell carcinoma, 21 with an adenocarcinoma and eight with other histology. Circulating antigens were measured by ELISA assay and the protein expression in primary tumors was analyzed by streptavidin-biotin immunohistochemical staining using specific monoclonal antibodies. The strong positivity for MMP-2 or MMP-9 in tumor predicted poor prognosis. The 5-year survival rates were 83 or 85% in patients negative for MMP-2 or MMP-9, respectively. Only 17% of the patients with a tumor highly positive for MMP-2 and 43% of those with a high positivity for MMP-9 survived at that time (Cox regression P=0.042 for MMP-2 and log rank P=0.046 for MMP-9). On the contrary, strong tissue positivity for TIMP-1 demonstrated a tendency for a favorable survival, although the difference did not reach statistical significance. In patients with a squamous cell carcinoma Stage I, low serum TIMP-1 (/=300ng/ml) associated with an increased survival rate, the 5-year survival being 81 versus 34% (log rank P=0.069) in patients with high or low serum levels for MMP-2/TIMP-2-complex, respectively. Tissue MMP-2 correlated to high expression of MMP-9 immunoreactive protein (P=0.003), but the serum levels of MMP-2/TIMP-2-complex or TIMP-1 did not correlate to the immunostaining of the corresponding tumors. We conclude that in lung carcinoma the best prognostic value is achieved by using immunohistochemistry for MMP-2 and MMP-9. In early disease, however, serum TIMP-1 or MMP-2/TIMP-2-complex could offer some further prognostic value.