Etiology of syringomyelia associated with Chiari type I malformation has been unknown. Moreover, the surgical procedure of foramen magnum decompression for this type of syringomyelia has not been standardized yet. No one procedure has been always successful, leading to many alternative procedures. The purpose of the present study is to elucidate pathway of cerebrospinal fluid into the syrinx cavity and to find out the best procedure for this disease. Fourty two patients with syringomyelia associated with Chiari type I malformation, which were diagnosed with magnetic resonance imaging (MRI), underwent surgical treatment. In all patients, craniocervical junction anomalies, cervical disc herniation and other spinal diseases were ruled out. There were 26 men and 16 women, ranging from 6 to 72 (mean: 42.3) years in age. The size, length and position of upper end of each syrinx cavity and the degree of the tonsillar herniation were measured on preoperative T1-weighted image and were compared each other. There were no significant relationship between the degree of tonsillar herniation and the size, length and position of syrinx cavity. No case showed that the upper end of syrinx cavity communicated to the 4th ventricle. The results suggest that the pathway of cerebrospinal fluid into the the syrinx cavity was not central canal from the 4th ventricle but microcanals in the spinal cord. All patients were carried out with foramen magnum decompression, which was divided into 4 groups according to the degree of decompression: 1) tonsillectomy group: 12 patients underwent subocciptital craniectomy (SOC) with patcy-graft dural plasty using lyophilized dura mater and tonsillectomy, 2) lysis group: 7 underwent SOC, dural plasty and microsurgical lysis of arachnoidal trabecula and fibrinoid filament around herniated tonsil, 3) plasty group: 17 underwent SOC and dural plasty and 4) dural group: 6 underwent SOC and removal of the outer layer of the dura mater. The mean follow-up periods were 3.7 years in tonsillectomy group, 3.6 years in lysis group, 2.3 years in plasty group, 1.8 years in dural group, respectively. Evaluation of the result following four types of surgical treatments was performed on clinical symptoms and the volume of syrinx cavity on sagittal MRI. The ratio in the area of the syrinx and spinal cord on preoperative and postoperative sagittal MRI were measured. There was no significant difference among 4 groups on the degree of reduction of syrinx in the sagittal plane as evaluated on MRI, whereas with regards to improvement of the clinical symptoms, dural group was significantly worse than the other three groups. The surgical procedure of dural plasty was clearly less invasive than those of tonsillectomy and lysis of subarachnoidal trabecula. These results suggest that we should select dural plasty as a primary surgical procedure for syringomyelia associated with Chiari type I malformation.