A theoretical framework for understanding acceleration-induced errors in phase-contrast magnetic resonance velocity measurements has been developed. An important result of this framework is the interpretation of acceleration-induced velocity errors as displacement artifacts due to the delay between velocity and spatial encoding. A rotating-disk phantom was used to confirm the theoretically predicted displacement times (the difference between theory and experiment was 8.2%). Errors were also observed in velocity profiles measured in regions of fluid acceleration downstream from a step stenosis. The magnitude of these errors could be predicted and corrected by using the analytic framework.