To assess the image quality and radiation dose of low-dose dual-source CT (DSCT) coronary angiography reconstructed using iterative reconstruction in image space (IRIS), in comparison with routine-dose CT using filtered back projection (FBP). Eighty-one patients underwent low-dose coronary DSCT using IRIS with two protocols: (a)100 kVp and 200 mAs per rotation for body mass index (BMI) < 25 (group I), (b)100 kVp and 320 mAs for BMI ≥ 25 (II). For comparison, two sex-and BMI-matched groups using standard protocols with FBP were selected: (a)100 kVp and 320 mAs for BMI < 25 (III), (b)120 kVp and 320 mAs for BMI ≥ 25 (IV). Image noise, signal to noise ratio (SNR) and modulation transfer function (MTF) 50% were objectively calculated. Two blinded readers then subjectively graded the image quality. Radiation dose was also measured. Image noise tended to be lower in IRIS of low-dose protocols: 22.0 ± 4.5 for group I versus 24.8 ± 4.0 for III (P < 0.001); 20.9 ± 4.5 for II versus 21.6 ± 4.9 for IV (P = 0.6). SNR was better with IRIS: 25.8 ± 4.4 for I versus 22.7 ± 4.6 for III (P < 0.001); 24.6 ± 5.4 for II versus 18.7 ± 4.5 for IV (P < 0.001). No differences in MTF 50% or image quality scores were seen between each two groups (P > 0.05). Radiation reduction was 40% for I and 51% for II, compared to standard protocols. Compared with routine-dose CT using FBP, low-dose coronary angiography using IRIS provides significant radiation reduction without impairment to image quality.