OBJECTIVES: To prospectively assess the impact of sinogram-affirmed iterative reconstruction (SAFIRE) on image quality of nonenhanced low-dose lung CT as compared to filtered back projection (FBP). METHODS: Nonenhanced low-dose chest CT (tube current-time product: 30 mAs) was performed on 30 patients at 100 kVp and on 30 patients at 80 kVp. Images were reconstructed with FBP and SAFIRE. Two blinded, independent readers measured image noise; two readers assessed image quality of normal anatomic lung structures on a five-point scale. Radiation dose parameters were recorded. RESULTS: Image noise in datasets reconstructed with FBP (57.4 ± 15.9) was significantly higher than with SAFIRE (31.7 ± 9.8, P < 0.001). Image quality was significantly superior with SAFIRE than with FBP (P < 0.01), without significant difference between FBP at 100 kVp and SAFIRE at 80 kVp (P = 0.68). Diagnostic image quality was present with FBP in 96% of images at 100 kVp and 88% at 80 kVp, and with SAFIRE in 100% at 100 kVp and 98% at 80 kVp. There were significantly more datasets with diagnostic image quality with SAFIRE than with FBP (P < 0.01). Mean CTDI(vol) and effective doses were 1.5 ± 0.7 mGy·cm and 0.7 ± 0.2 mSv at 100 kVp, and 1.4 ± 2.8 mGy·cm and 0.5 ± 0.2 mSv at 80 kVp (P < 0.001, both). CONCLUSIONS: Use of SAFIRE in low-dose lung CT reduces noise, improves image quality, and renders more studies diagnostic as compared to FBP. KEY POINTS : • Low-dose computed tomography is an important thoracic investigation tool. • Radiation dose can be less than 1 mSv with iterative reconstructions. • Iterative reconstructions render more low-dose lung CTs diagnostic compared to conventional reconstructions.