To evaluate the effect of radiation dose reduction on image quality and diagnostic accuracy of coronary computed tomography (CT) angiography. Coronary CT angiography studies of 40 patients with (n = 20) and without (n = 20) significant (≥50 %) stenosis were included (26 male, 14 female, 57 ± 11 years). In addition to the original clinical reconstruction (100 % dose), simulated images were created that correspond to 50, 25 and 12.5 % of the original dose. Image quality and diagnostic performance in identifying significant stenosis were determined. Receiver-operator-characteristics analysis was used to assess diagnostic accuracy at different dose levels. The identification of patients with significant stenosis decreased consistently at doses of 50, 25 and 12.5 of the regular clinical acquisition (100 %). The effect was relatively weak at 50 % dose, and was strong at dose levels of 25 and 12.5 %. At lower doses a steady increase was observed for false negative findings. The number of coronary artery segments that were rated as diagnostic decreased gradually with dose, this was most prominent for smaller segments. The area-under-the-curve (AUC) was 0.90 (p = 0.4) at 50 % dose; accuracy decreased significantly with 25 % (AUC 0.70) and 12.5 % dose (AUC 0.60) (p < 0.0001), with underestimation of patients having significant stenosis. The clinical acquisition protocol for evaluation of coronary artery stenosis with CT angiography represents a good balance between image quality and patient dose. A potential for a modest (<50 %) reduction of tube current might exist. However, more substantial reduction of tube current will reduce diagnostic performance of coronary CT angiography substantially.