This study aimed to investigate the impact of conventional linear accelerator errors in the gantry angle and multi-leaf collimators (MLCs) on the intensity-modulated arc therapy (IMAT) radiation dose for brain tumors. Seven typical planning datasets of patients with brain tumors were randomly selected and retreated using the IMAT conventional linear accelerator Varian 23EX. Random and system errors were introduced to modify these plans while maintaining the remaining treatment information. The dose-volume parameters of the target and critical organs were compared and analyzed. The results showed that for random single-sided MLCs errors, if a group of errors was moved outward by 2 mm, the planning target volume (PTV)-D_mean increased by approximately 2.7%. If one side of the MLCs expanded outward with MLC-1, MLC-2, and MLC-rnd, the mean deviations of PTV-D_mean were 1.12%, 2.15%, and 1.15%, respectively, and the D_mean deviations of the lenses were 5.25%, 9.97%, and 5.49%, respectively. The maximum deviation of PTV-D_mean because of the deviation of two sets of leaves caused by system errors (±2 mm) was 0.8%, and maximum dose and D_mean of the lens and brainstem increased by 11.4%, 15.8% and 1.99%, 1.07%, respectively. By introducing a gantry angle error, the maximum doses of PTV-D_min and-D_mean reduced by 2.75% and 0.35%, respectively. An incorrect gantry position results in significant changes in the dosimetric deviation of the OAR dose-volume histogram, especially for the lens. Note that the maximum and mean dose deviations of a patient’s lens if the gantry deviation (–2°) were 16.9% and 38.5% higher than those of the reference plan, respectively. The dose deviation caused by random errors of the MLCs leaves was influenced by the actual size of the random displacement errors of the leaves and weight of the beam subfield at a specific arc segment. The maximum dose deviation occurs if MLCs move to the maximum simulation error (±2 mm). The gantry rotation error had no significant impact on the target area PTV dose distribution; however, it had a significant dose deviation for the OAR.