ObjectiveTo investigate the impact of different dose algorithms and calculation angle intervals (DCAI) on the in vivo dose (IVD) verification of small-field arc therapy in stereotactic body radiation therapy (SBRT).MethodsWe made an exit-dose-measuring and positioning device (EDPD) for the SRS MapCHECK (SMC) using polymethyl methacrylate (PMMA). Computed tomography data for the anthropomorphic head phantom, SMC, and EDPD combination were acquired with 1 mm slice thickness and spacing. SBRT partial arc plans were created using an SBRT cone, block, and a small square open field, with a gantry rotation angle of 60°. The dose distribution was calculated using three different dose algorithms [Pencil Beam (PB), CC Convolution (CCC), and Monte Carlo (MC)], with 1 mm isotropic resolution. We also used three different DCAIs (1°, 3°, 5°) with the PB and CCC algorithms to calculate the dose distribution of each plan three times. The uncertainty of each control point for the MC algorithm was set to 1%. The SMC was used to measure the exit dose outside the phantom for IVD verification, the detector plane was located 182.5 mm outside the scan center.ResultsWithin the phantom, the minimum passing rate of 3D gamma analysis (1%/1 mm) for the dose distributions calculated at different DCAIs was 99.1%, and the maximum relative deviation (RD) of the central point dose (CPD) was <0.2%. The average RD of the CPD for IVD verification was about 30% (range 16.71%–50.0%) for PB; -0.36% ± 1.82% (1° DCAI), -3.18% ± 7.83% (3° DCAI), and 3.69% ± 11.56% (1° DCAI) for CCC; and -0.38% ± 0.76 for the MC algorithm. The passing rates of 2D gamma analysis (3%/3 mm) between the predicted exit dose and the IVD were 100% for MC and >90% for the CCC algorithm at 1° DCAI.ConclusionThe DCAI for exit-dose calculations should be ≤1° using the CCC algorithm. Furthermore, among the three algorithms verified in the current study, the MC algorithm showed the highest accuracy, followed by CCC, with the PB algorithm having the worst performance. The PB algorithm is thus not suitable for exit-dose calculation or IVD verification of SBRT.