Abstract: In this paper we give an overview over a series of experiments to visualize and measure flow fields in the human vascular system with respect to their diagnostic capabilities. The experiments utilize a selection of GPU-based sparse and dense flow visualization algorithms to show the diagnostic opportunities for volumetric cardiovascular phase contrast magnetic resonance imaging data sets. Besides classical hardware accelerated particle and line-based approaches, an extensible tublet-based visualization, a four-dimensional volumetric line integral convolution and a new two-dimensional cutting plane tool for three-dimensional velocity data sets have been implemented. To evaluate the results, several hearts of human subjects have been investigated and a flow phantom was built to artificially simulate distinctive flow features. Our results demonstrate that we are able to provide an interactive tool for cardiovascular diagnostics with complementary hardware accelerated visualizations.