DM-Ice is an experiment designed for a direct detection search for dark matter. Using a NaI(Tl) target, DM-Ice searches for WIMP (weakly interacting massive particles) dark matter via scintillation associated with nuclear recoil in the crystal, which is then observed by PMTs. DM-Ice can test the DAMA/LIBRA result, using the same target material while running in the Southern Hemisphere. The DM-Ice prototype runs at the South Pole station, deployed underneath the IceCube Neutrino Observatory. This thesis describes the simulation work performed in order to understand the prototype detector, DM-Ice17. Dark matter background, evidence, and current understanding are discussed by way of introduction to the field. Description and discussion of detection methods and current experimental dark matter detection results follows. The DM-Ice detector itself is then considered in detail, in terms of motivation, design, and function. The assembly, deployment and operation of DM-Ice17 is also discussed. The purpose of simulating the radioactive backgrounds present in the DM-Ice17 detector is to understand the detector and the contamination levels present in each of its components, and to provide information needed for design and material selection for the full-scale DM-Ice detector. The Geant4 simulation toolkit was used to simulate the detector. The simulation is described in terms of geometry, particle decay and propagation, and producing an energy spectrum. This simulated energy spectrum was then used to characterize the detector, and this process is described as well. This thesis demonstrates that the simulation I have created aligns well with the data from DM-Ice17. This simulation allows insight into and verification of the radioactive contamination of each of the component of the detector, as well as that of its surroundings. The simulation also allows for detailed consideration of the contamination levels in different materials, which is needed in order to select materials and designs for the full-scale DM-Ice detector. Details regarding contributions of different isotopes in each region of the detector to the region of interest (low-energy; approximately 0-10 keV) are extracted from the simulation, which allows optimization of understanding what degree of cleanliness is needed for purposes of our dark matter search.