Purpose: Iterative metal artifact reduction (IMAR) is new commercially available algorithm based on normalized iterative interpolation. This study was to assess the performance of IMAR compared to filtered back projection (FBP) reconstruction on metal artifacts reduction,using thoracic phantom with pedicle screws implanted and combined with clinical cases. Methods: ①Images were acquired on Siemens Perspective CT (scan parameters: 130 kV, CAREDose4D with 80 reference mAs). Two pedicle screws were inserted into the anthropomorphic phantom (Lungman N1) on two sides of the fifth thoracic vertebra. The phantom was scanned with and without the screws respectively. Images without metal artifacts were reconstructed with FBP. Images with metal artifacts were reconstructed with FBP and IMAR respectively. Three regions of interest (ROIs) were selected on tissues including aorta, pulmonary and paravertebral soft tissue on image slice adjacent to the screws. The CT numbers and standard deviations (noise) over ROIs were measured. The deviation of CT number was calculated as the difference between CT values in images with and without metal artifacts. The noise and deviation of CT number were compared between FBP and IMAR. ②26 cases that received chest CT examination and with pedicle screw implant in their scanning range were collected in our hospital. The scanning parameters and image reconstruction methods were the same with before, and two skilled radiologists independently evaluated the image quality of reconstructed image. Results: ①In the phantom study, the noise were significantly reduced by IMAR compared to FBP ,in all the three ROIs. The deviation of CT number was significantly smaller with IMAR compared to that with FBP,in all the three ROIs . ② The subjective evaluation scores of the two image reconstruction methods were (1.00±0.00)and(2.83±0.51), and the results showed a statistically significant difference (P = 0.00). Conclusion: IMAR significantly reduced streak artifacts of metal implant in thoracic phantom, and adjusted the CT values of artifact affected tissues through normalized iterative interpolation. IMAR significantly improved the image quality of thoracic phantom CT with metal implant.