Background About 50%-70% of patients with Chiari malformation I(CMI)presented with syringomyelia(SM),whichis supposed to be related to abnormal cerebrospinal fluid(CSF)flow around the foramen magnum.The aim of this studywas to investigate the cerebrospinal fluid dynamics at levels of the aqueduct and upper cervical spine in patients withCMI associated with SM,and to discuss the possible mechanism of formation of SM.Methods From January to April 2004,we examined 10 adult patients with symptomatic CMI associated with SM and 10healthy volunteers by phase-contrast MRI.CSF flow patterns were evaluated at seven regions of interest(ROI):theaqueduct and ventral and dorsal subarachnoid spaces of the spine at levels of the cerebellar tonsil,C2-3,and C5-6.TheCSF flow waveforms were analyzed by measuring CSF circulation time,durations and maximum velocities of cranial-andcaudal-directed flows,and the ratio between the two maximum velocities.Data were analyzed by ttest using SPSS 11.5.Results We found no definite communication between the fourth ventricle and syringomyelia by MRI in the 10 patients.In both the groups,we observed cranial-directed flow of CSF in the early cardiac systolic phase,which changed thedirection from cranial to caudal from the middle systolic phase to the early diastolic phase,and then turned back in cranialdirection in the late diastolic phase.The CSF flow disappeared at the dorsal ROI at the level of C2-3 in 3 patients and 1volunteer,and at the level of C5-6 in 6 patients and 3 volunteers.The durations of CSF circulation at all the ROIs weresignificantly shorter in the patients than those in the healthy volunteers(P=0.014 at the midbrain aqueduct,P=0.019 atthe inferior margin of the cerebellar tonsil,P=0.014 at the level of C2-3,and P=0.022 at the level of C5-6).No significantdifference existed between the two groups in the initial point and duration of the caudal-directed CSF flow during acardiac cycle at all the ROIs.The maximum velocities of both cranial-and caudal-directed CSF flows were significantlyhigher in the patients than those in the volunteers at the aqueduct(P=0.018 and P=0.007)and ventral ROI at the inferiormargin of the cerebellar tonsil(P<0.001 and P=0.002),as so did the maximum velocities of the caudal-directed flow in theventral and dorsal ROIs at the level of C2-3(P=0.004;P=0.007).Conclusions The direction of CSF flow changes in accordance with cardiac cycle.The syringomyelia in patients withCMI may be due to the decreased circulation time and abnormal dynamics of the CSF in the upper cervical segment.Thedecompression of the foramen magnum with dural plasty is an alternative for patients with CMI associated with SM. Background About 50% -70% of patients with Chiari malformation I (CMI) presented with syringomyelia (SM), whichis supposed to be related to abnormal cerebrospinal fluid (CSF) flow around the foramen magnum. The aim of this study was to investigate the cerebrospinal fluid dynamics at levels of the aqueduct and upper cervical spine in patients with CMI associated with SM, and to discuss the possible mechanism of formation of SM. Methods From January to April 2004, we examined 10 adult patients with symptomatic CMI associated with SM and 10 healthy volunteers by phase-contrast MRI. CSF flow patterns were evaluated at seven regions of interest (ROI): the aqueduct and ventral and dorsal subarachnoid spaces of the spine at levels of the cerebellar tonsil, C2-3, and C5-6. TheCSF flow waveforms were analyzed by measuring CSF circulation time, durations and maximum velocities of cranial-andcaudal-directed flows, and the ratio between the two maximum velocities. Data were analyzed by ttest using SPSS 11.5. Results We fo und no definite communication between the fourth ventricle and syringomyelia by MRI in the 10 patients. In both the groups, we observed cranial-directed flow of CSF in the early cardiac systolic phase, which changed the direction from cranial to caudal from the middle systolic phase to the early diastolic phase, and then turned back in cranial direction in the late diastolic phase. The CSF flow disappeared at the dorsal ROI at the level of C2-3 in 3 patients and 1 volunteer, and at the level of C5-6 in 6 patients and 3 volunteers. Durations of CSF circulation at all the ROIs weresignificantly shorter in the patients than those in the healthy volunteers (P = 0.014 at the midbrain aqueduct, P = 0.019 atthe inferior margin of the cerebellar tonsil, P = 0.014 at the level of C2-3, and P = 0.022 at the level of C5-6) .No significant difference between the two groups in the initial point and duration of the caudal-directed CSF flow during acardiac cycle at all the ROIs.The maximum velocities of both cranial-and caudal-directed CSF flows were significantlyhigher in the patients than those in the volunteers at the aqueduct (P = 0.018 and P = 0.007) and ventral ROI at the inferiormargin of the cerebellar tonsil (P <0.001 and P = 0.002) as so did the maximum velocities of the caudal-directed flow in the event and dorsal ROIs at the level of C2-3 (P = 0.004; P = 0.007) .Conclusions The direction of CSF flow changes in accordance with cardiac cycle. The syringomyelia in patients withCMI may be due to the decreased circulation time and abnormal dynamics of the CSF in the upper cervical segment.Thecompression of the foramen magnum with dural plasty is an alternative for patients with CMI associated with SM.