The cerebrospinal fluid flow in the third ventricle of the brain and the aqueduct of Sylvius was studied using computational fluid dynamics (CFD) based on subject-specific boundary conditions derived from magnetic resonance imaging (MRI) scans. The flow domain geometry was reconstructed from anatomical MRI scans by manual image segmentation. The movement of the domain boundary was derived from MRI brain motion scans. Velocimetric MRI scans were used to reconstruct the velocity field at the inferior end of the aqueduct of Sylvius based on the theory of pulsatile flow in pipes. A constant pressure boundary condition was assigned at the foramina of Monro. Three main flow features were observed: a fluid jet emerging from the aqueduct of Sylvius, a moderately mobile recirculation zone above the jet and a mobile recirculation below the jet. The flow in the entire domain was laminar with a maximum Reynolds number of 340 in the aqueduct. The findings demonstrate that by combining MRI scans and CFD simulations, subject-specific detailed quantitative information of the flow field in the third ventricle and the aqueduct of Sylvius can be obtained.