The recently introduced coronary artery disease reporting and data system (CAD-RADS) evaluated by computed tomography and based on stenosis severity, might not adequately reflect the complexity of CAD. We explored the relationship between CAD-RADS and the spatial distribution, burden, and complexity of lesions by invasive coronary angiography (ICA). Stable patients who underwent coronary computed tomography angiography (CCTA) and ICA comprised the study population. Patients were classified according to the CAD-RADS: 0, No plaque; 1, 1–24% stenosis; 2, 25–49%; 3, 50–69%; 4A, 70–99%; 4B, left main stenosis or 3-vessel obstructive disease; and 5, total occlusion. Based on ICA findings, we calculated the SYNTAX score and the CAD extension index. Ninety-one patients were included, with a mean age of 61.4 ± 10.5 years (74% male). We found significant relationships between CAD-RADS and both the SYNTAX score (p < 0.0001) and the CAD extension index (p < 0.0001), although the complexity of coronary anatomy differed among patients with CAD-RADS ≥ 4A. Among patients with CAD-RADS < 4, the mean segment involvement score (SIS) was 8.4 ± 4.0, 52% of them with a SIS > 5. Of the 30 patients with CAD-RADS 5, 9 (30%) affected distal segments or secondary branches, and 9 (30%) had concomitant severe non-extensive disease at ICA. Regarding the spatial distribution of the non-occluded most severe lesions, 27 (44%) comprised distal segments or secondary branches. In the present study including a high-risk population, we identified diverse coronary anatomy complexity scenarios and relevant differences in spatial distribution sharing the same CAD-RADS classification.