We explore the feasibility of coronary calcium subtraction computed tomography angiography (CCTA) in patients with high calcium scores using invasive coronary angiography as the gold standard. Eleven patients with calcium scores of >400 underwent CCTA using a subtraction protocol followed by invasive coronary angiography. In addition to standard reconstructions, subtracted images were obtained using a dedicated subtraction algorithm. A total of 55 calcified segments were evaluated for image quality [using a 4-point scale ranging from 1 (uninterpretable) to 4 (good)] and the presence of significant (≥50 %) luminal stenosis. Conventional and subtracted CCTA were compared using quantitative coronary angiography (QCA) as the gold standard. The average image quality of conventional CCTA was 2.5 ± 0.6 versus 3.1 ± 0.6 on subtraction CCTA (P < 0.001). The percentage of segments with a score 1 or 2 was reduced from 41.8 to 12.7 % after coronary calcium subtraction (P = 0.002). On QCA, significant stenosis was observed in 16 segments. The area under the receiver operating characteristics curve to detect ≥50 % stenosis on QCA increased from 0.741 [95 % confidence interval (CI) 0.598–0.885] for conventional CCTA to 0.905 (95 % CI 0.791–1.000) for subtraction CCTA (P = 0.003). In patients with extensive calcifications undergoing CCTA, coronary calcium subtraction may improve the evaluation of calcified segments.