Purpose: Dynamic contrast-enhanced (DCE) imaging of the liver with high temporal resolution is a method of measuring perfusional changes in the liver and in focal liver lesions. Several studies have shown the utility of perfusion metrics in diagnosis of advanced liver fibrosis, and for characterization of angiogenic activity of hepatocellular carcinoma and liver metastases, as well as for monitoring antiangiogenic treatment response (1-3). 2D or 3D interpolated spoiled gradientecho sequences are routinely used for liver perfusion imaging (PWI). Compared to 2D, 3D acquisitions have the advantage of covering the entire liver, which is essential when evaluating patients with multiple liver lesions, and in liver fibrosis and cirrhosis, where there is a heterogeneous distribution of the disease. However, this extended anatomic coverage is at the expense of lower temporal resolution and/or lower in-plane spatial resolution. Furthermore, prior to performing DCE analysis coregistration of dynamic volumes, which can be cumbersome and technically challenging, is essential to compensate for respiratory motion. For PWI to gain acceptance in routine clinical practice, acquisition schemes need to be robust to respiratory motion, and should be able to acquire simultaneous high spatial and temporal resolution T1-weighted dynamic data with whole-liver coverage. One approach for accelerated imaging acquisition is the use of compressed sensing (CS) reconstruction techniques, where temporal correlations in the dynamic imaging data are exploited to reduce k-space sampling necessary to generate an image (4, 5). Radial k-space sampling schemes are well suited for CS due to the presence of inherent incoherent aliasing artifacts, particular when using the golden-angle scheme (6). Recently we developed a novel reconstruction method that combines compressed sensing and parallel imaging for radial trajectories (k-t RASPS: RAdial SParse-Sense). Therefore, the purpose of this study was to assess the feasibility of performing free-breathing high spatial and temporal resolution imaging of the whole liver with radial golden-angle sampling and k-t RASPS reconstruction (GA-CS). Furthermore, our aim was to compare the signal-intensity-time curve of the GACS reconstructions to the regridded (RG) data sets, to investigate if temporal blurring effects occur in the CS reconstructions.