PURPOSE: Dynamic contrast-enhanced imaging provides unique physiological information, notably the endothelial permeability (K(trans) ), and may improve the diagnosis and management of multiple pathologies. Current acquisition methods provide limited spatial-temporal resolution and field-of-view, often preventing characterization of the entire pathology and precluding measurement of the arterial input function. We present a method for highly accelerated dynamic imaging and demonstrate its utility for dynamic contrast-enhanced modeling. METHODS: We propose a novel Poisson ellipsoid sampling scheme and enforce multiple spatial and temporal l1 -norm constraints during image reconstruction. Retrospective and prospective analyses were performed to validate the approach. RESULTS: Retrospectively, no mean bias or diverging trend was observed as the acceleration rate was increased from 3× to 18×; less than 10% error was measured in K(trans) at any individual rates in this range. Prospectively accelerated images at a rate of 36× enabled full brain coverage with 0.94 × 0.94 × 1.9 mm(3) spatial and 4.1 s temporal resolutions. Images showed no visible degradation and provided accurate K(trans) values when compared to a clinical population. CONCLUSION: Highly accelerated dynamic MRI using compressed sensing and parallel imaging provides accurate permeability modeling and enables full brain, high resolution acquisitions. Magn Reson Med 000:000-000, 2013. © 2013 Wiley Periodicals, Inc.