The fundamental goal of radiation oncology is the accurate delivery of radiation dose to the clinically defined tumor volume while minimizing dose to healthy tissue. Stereotactic radiosurgery is a specialized treatment technique most often used to deliver a highly conformal radiation dose to a well-localized volume within the brain. One can perform stereotactic radiosurgery on a linear accelerator through the use of secondary collimators to shape the beam and achieve a high dose falloff, thus sparing adjacent critical structures. During treatment, the patient receives a number of external beam arcs entering the body along different planes. In the radiation oncology department at Our Lady of Lourdes Hospital in Binghamton, NY, dosimetrists and physicists use XiO software provided by Computerized Medical Systems to create radiotherapy treatment plans. Radiation therapists then deliver these plans on a Varian linear accelerator under the supervision of both a doctor and a medical physicist. In this paper we discuss the dose accuracy of our previously commissioned convolution algorithm in stereotactic radiosurgery planning. Radiation oncology centers do not traditionally use the convolution algorithm to generate these plans, choosing instead to use empirical data to model the small-diameter beams. We focus on three stereotactic cone sizes: 14, 20, and 30 mm. First we explore the inadequacies of the Sun Nuclear MapCHECK device in analyzing stereotactic beam fluence. We then capture percent depth dose data using an IBA Dosimetry CC01 microchamber and Blue Phantom to determine the convolution algorithm’s overestimation of PDD values at depths exceeding 10 cm. We next make use of Kodak EDR2 film dosimetry to analyze beam profiles. We determine that there is a general broadening of the modeled beams, as shown by deviations in both FWHM and penumbra values for the three stereotactic cones. Based on our results, we conclude that the CMS convolution algorithm as commissioned for intensity modulated radiation therapy treatments does not accurately replicate a clinically viable stereotactic radiation treatment plan. We discuss ways to improve the accuracy of the algorithm, including modifications to both the beam model’s energy spectrum and penumbra focal radiation block sigma.