Abstract. A long-standing challenge in the field of volume rendering is to obtain high quality images in near real-time. This is particularly important for scientific datasets where highly precise results are required to ensure accurate data interpretation. This paper presents RZSweep as a new hardware-assisted technique for volume rendering of regular datasets based on the plane-sweep paradigm. Although some research has been done in the past for irregular datasets, this paper presents the first attempt to explore the capabilities of the plane sweep paradigm for regular datasets. RZSweep is an exact, object order, direct volume rendering (DVR), back-to-front projection algorithm wherein a plane sweeps through the volume in depth order, projecting all the data elements within the user specified threshold onto the image plane. The algorithm uses a hardware rendering pipeline to composite the final image. The space complexity of the algorithm is bound to the number of vertices in the largest diagonal plane of the dataset. A prototype of the algorithm renders a dataset of size 2563 in near real-time of 0.77 seconds on a single off-the-shelf commodity hardware PC. Every data element within the specified threshold contributes to the final image in its correct spatial order that is guaranteed by the use of a heap.