3D-T1ρ mapping sequences are useful MRI methods in various neuropathologies but its data acquisition requires long scan times. We compared the performance of 5 compressive sensing (CS) algorithms with acceleration factors (AF) up to 10. We evaluated image quality and T1ρ estimation errors as a function of AF. Six healthy volunteers were recruited and they underwent T1ρ imaging of the whole brain with full Cartesian acquisition. Assessment of image reconstruction and T1ρ estimation errors in this study show that the CS method using spatial and temporal finite differences as a regularization function performs the best for accelerating T1ρ quantification in the brain.