OBJECTIVE: To establish a rabbit model of cerebral spinal flow metastasis, to analyze the growth rate of tumor, and to investigate the value of MRI in monitoring the biology of tumor compared with pathology. METHODS: Twenty-four New Zealand white rabbits were inoculated with suspension of VX(2) tumor cells in the subarachnoid space via the foramen magnum (experimental group), and 6 rabbits were inoculated with normal saline (control group). MRI examination, including non-enhanced T(1)WI, T(2)WI, and FLAIR sequences and then T(1)WI, FLAIR after dynamic contrast enhanced with Gd-DTPA were done 7 approximately 22 days after inoculation with a 3-day interval. The rabbits were killed after the last MRI scan with their spinal cords, spinal meninges, and tumor taken out to undergo microscopy. RESULTS: (1) MRI plain scan showed that in the experimental group 2 nodi in the medulla and 1 nodes in the cervical spinal cord were found with low signal on T(1)WI and high signal on T(2)WI; and FLAIR imaging showed local lesions with medial signal in 6 rabbits (25%). And no abnormal signs were seen in the control group. (2) MRI enhancement showed that in the experimental group the images of 15 rabbit models were enhanced markedly with irregular thickening of meninges or nodules at the subarachnoid space on T(1)WI, positive signs were confirmed on FLAIR sequence in 16 of the 24 rabbits, and positive signs were noted on DCE-MRI scanning in 18 of the 24 rabbits (75%). In the control group 5 of the 6 rabbits were negative in images. Microscopy showed thickened of meninges and spinal meninges in 20 of the 24 rabbits of the experimental group and spinal cord metastasis in 22 rabbits. No pathological changes were seen in the control group. Statistics showed a CSF metastasis rate of 91.67%. There were significant difference between the plain scan and T(1)WI with enhancement (P < 0.01) and between FLAIR scan and FLAIR enhancement scans. There was a significant difference between T(1)WI and FLAIR enhancement and pathological findings (P < 0.05). There was no significant difference between DCE-MRI method and pathological results (P > 0.05). CONCLUSION: Gd-DTPA enhanced MRI scan sequences has a high sensitivity and specificity and can be used in monitoring the growth of CSF metastasis. There is a disparity between the MRI signs and pathological findings. It is a key that to improve the spatial resolution of machine and to investigate the best method for detecting early metastasis.