Purpose: Cerebral metastases represent the most common type of brain tumors. This study investigated the effects of endogenous endostatin on hematogenous cerebral melanoma metastases. Experimental Design: Murine K1735 melanoma cells were transfected with the mouse endostatin cDNA. Experimental tumors were induced either by s.c. injection, intracerebral implantation, or via injection into the internal carotid artery to simulate hematogenous metastatic spread. The effects of endostatin expression on tumor incidence, growth pattern, and vascularity were analyzed. Results: In vitro secretion of endostatin by 2.5 × 10⁵ cells within 24 hours was 0.12 ± 0.03 ng, 4.35 ± 0.4, and 1.18 ± 0.7 ng/mL for wild type and two endostatin-transfected K1735 clones termed K1735-endo/2 and K1735-endo/8, respectively. Tumor inhibition in vivo correlated with endogenous endostatin production. Within 25 days, growth of s.c. K1735-endo/2 tumors was <20% compared with wildtype controls. Following intracerebral implantation the average survival time of mice was 27.8 ± 2.6 versus 13.3 ± 3.7 days in the K1735-endo/2 versus the wild-type group, respectively. Intracarotid injection of 1 × 10⁵ wild-type cells killed the mice within 24 ± 1.8 days. In contrast, endostatin expression prevented macroscopic metastatic tumor growth in 11 of 12 mice, although viable microscopic tumor pockets were detectable in all animals. Conclusion: Endostatin inhibits tumor progression of multiple cerebral metastases in vivo. Hematogenous micrometastases are more efficiently suppressed than tumors resulting from high focal cell numbers which may be due to a higher angiogenic signaling exerted by massive cell deposits. Endostatin may prevent solid tumor growth more effectively by inhibition of early angiogenesis.