Background: Intensity-modulated radiotherapy (IMRT) techniques have a steep dose distribution, leading to large dosimetric errors caused by incorrect daily patient setup. Image-guided radiotherapy (IGRT) is an essential technique to verify accurate patient setup. The accuracy of verifying a patient’s position depends on many factors including the image guidance system and image registration software. Objectives: This study investigated the dosimetric and geometric differences between the original plan and simulated plans with setup errors associated with kilovoltage cone-beam computed tomography (kV-CBCT), kV, kV planar image, and electronic portal imaging detector (EPID) for brain tumor treatment using IMRT. Materials and methods: A PIXY Anthropomorphic Training/Teaching Phantom was used in this study. IMRT treatment plans with five co-planar fields from a 6 MV Elekta Versa HD linear accelerator were generated using the RayStation computer treatment planning system. Three image guidance systems, including kV-CBCT, kV planar imager, and EPID were used to perform image registration. To evaluate the efficiency of each image guidance system, a simulated setup error by couch was shifted 0, ±2, and ±4 mm in the lateral, longitudinal, and vertical planes. Errors in the image registration from the actual couch shift were collected. Measured errors were used to generate the treatment plans of 25 fractions using the IMRT technique by random shifts of the isocenter of each fraction while maintaining other planning parameters of the original plan. The dose deviation in planning target volume (PTV) and geometric deviation compared to original plan were recalculated and analyzed. Results: Accuracy of image registration in all image guidance systems indicated that the registration errors were less than 1.7, 2.0, and 1.0 mm for the kV-CBCT, kV planar, and EPID, respectively. The average PTV dosimetric deviation induced by the setup error ranged from -2 to 2 mm per fraction; this study showed dosimetric deviation at D98% approximately -6%, D95% approximately -4% and D2% below -1%, while the average of isodose distribution shift of -5% inside the PTV of the isocenter region were -1.58±0.67, -1.28±0.52 and -1.30±0.80 mm in the cerebella, parasagittal, and convex areas, respectively. An isodose shift of -10% was <1 mm in all PTV locations. Conclusion: The efficiency of image guidance resulted in small errors within ±2 mm using kV-CBCT, kV planar, and EPID. The setup error influenced daily dose distribution, while the PTV recorded underdose of about 4% in the brain IMRT technique.