Objective To evaluate the effect of setup errors upon the target area and the organs at risk (OAR) during radiotherapy for prostate cancer. Methods Twelve prostate cancer patients receiving treatment in the recent 1 year were randomly recruited in this study. The position of each patient was verified by using cone beam CT (CBCT) for 6-10 times during the treatment. In treatment planning system (TPS), the isocenter position was moved along the setup errors with averaging error value (Plan_A) and each CBCT value (Plan_F). The dose distribution was recalculated without changing the beam setting, weight factors and monitor units (MUs). The dose difference was statistically compared between the simulation and original plans (Plan_O). Results For clinical target volume (CTV) D95, there was a significant difference between Plan_A and Plan_O (P=0.008), whereas no significant difference was observed between Plan_F and Plan_O. There were significant differences between Plan_F and Plan_O, Plan_A and Plan_O (P=0.004, and 0.041) for the planned target volume (PTV) D95.For OAR, rectal V60, Dmax, left femoral V20, Dmax and right femoral Dmax significantly differed between Plan_F and Plan_O (P=0.026, 0.015, 0.041, 0.049, 0.003). However, only left femoral Dmax significantly differed between Plan_A and Plan_O (P=0.045). The movement in the superior-inferior (SI) direction was significantly correlated with the changes in the rectal V40, V50 and V60 and PTV D95(r=-0.785, -0.887, -0.833, 0.682). The movement in the anterior-posterior (AP) direction was significantly associated with the variations in the bladder V20, V30, V40, V50 and V60(r=-0.945, -0.823, -0.853, -0.818, -0.774). The evaluation indexes of all normal tissues in the re-plan could meet the clinical requirements. However, the volume of target prescription volume had different levels of deficit, and the deficit of Plan_F was greater than that of Plan_A. Conclusions The simulation results of averaging into the TPS underestimates the effect of daily setup errors on the dose distribution. The effect of setup errors on the dose distribution in target area is greater than that of normal tissues. Y-direction errors are more likely to cause the variations of the rectal and PTV dose, and the errors in the z-direction are inclined to cause the changes in the bladder dose. Key words: Prostate neoplasm/radiotherapy; Setup error; Dose distribution