An approach to efficiently measure three-dimensional velocity vector fields and turbulent kinetic energy of blood flow is presented. Multipoint phase-contrast imaging is used in combination with Bayesian analysis to map both mean and fluctuating velocities over a large dynamic range and for practically relevant signal-to-noise ratios. It is demonstrated that the approach permits significant spatiotemporal undersampling to allow for clinically acceptable scan times. Using numerical simulations and in vitro measurements in aortic valve phantoms, it is shown that for given scan time, Bayesian multipoint velocity encoding provides consistently lower errors of velocity and turbulent kinetic energy over a larger dynamic range relative to previous methods. In vitro, significant differences in both peak velocity and turbulent kinetic energy between the aortic CoreValve (150 cm/s, 293 J/m(3) ) and the St. Jude Medical mechanical valve (120 cm/s, 149 J/m(3) ) were found. Comparison of peak turbulent kinetic energy measured in a patient with aortic stenosis (950 J/m(3) ) and in a patient with an implanted aortic CoreValve (540 J/m(3) ) revealed considerable differences relative to the values detected in healthy subjects (149 ± 12 J/m(3) ) indicating the potential of the method to provide a comprehensive hemodynamic assessment of valve performance in vivo.