Our previous clinicopathologic study revealed an inverse association of liver metastasis of colorectal cancer and stromal expression of matrix metalloproteinase-9 (MMP-9) or urokinase receptor (uPAR). This suggests that host cells, particularly macrophages, expressing matrix-degrading enzymes/factors could be protective for the host against hematogenous metastasis. However, our previous study was unable to differentiate whether our results were causes or effects of widely spread cancer. To solve this point, we designed the present study on colorectal cancers that developed hematogenous metastasis after operation, ie., metachronous hematogenous metastasis. These cancers, being solely micrometastasized at the time of operation, allowed us to eliminate possible systemic effects by widely spread cancer. Sixty-two primary tumors with metachronous metastasis showed a decreased number of MMP-9+ stromal cells and CD68+ macrophages along the invasive margin with unchanged uPAR+ stromal area as compared with those in 72 control cases, which were free from tumor metastasis or recurrence for more than 5 years. Therefore, we judged the decrease of MMP-9+ host cells or macrophages in the primary site is irrelevant of effects of widely spread metastasis but probably related to causes of metastasis. Our data also characterized the metachronous metastasis group by uPAR expression in fibroblasts. The number of uPAR+ cancer cells, although small in number, were also larger in the metachronous metastasis group. Our data revealed that macrophages, a major source of uPAR and one of the sources of MMP-9, could be inhibitory to hematogenous metastasis, while uPAR+ fibroblasts and cancer cells, in turn, facilitate hematogenous metastasis. This suggests the functional multiplicity of matrix degradation processes in cancer tissue.