Characterization of endothelial shear stress (ESS) may allow for prediction of the progression of atherosclerosis. The aim of this investigation was to develop a non-invasive approach for in vivo assessment of ESS by coronary computed tomography angiography (CTA) and to compare it with ESS derived from invasive coronary angiography (ICA). A total of 41 patients with mild or intermediate coronary stenoses who underwent both CTA and ICA were included in the analysis. Two geometrical models of the interrogated vessels were reconstructed separately from CTA and ICA images. Subsequently, computational fluid dynamics were applied to calculate the ESS, from which ESS_CTA and ESS_ICA were derived, respectively. Comparisons between ESS_CTA and ESS_ICA were performed on 163 segments of 57 vessels in the CTA and ICA models. ESS_CTA and ESS_ICA were similar: mean ESS: 4.97 (4.37–5.57) Pascal versus 4.86 (4.27–5.44) Pascal, p = 0.58; minimal ESS: 0.86 (0.67–1.05) Pascal versus 0.79 (0.63–0.95) Pascal, p = 0.37; and maximal ESS: 14.50 (12.62–16.38) Pascal versus 13.76 (11.44–16.08) Pascal, p = 0.44. Good correlations between the ESS_CTA and the ESS_ICA were observed for the mean (r = 0.75, p < 0.001), minimal (r = 0.61, p < 0.001), and maximal (r = 0.62, p < 0.001) ESS values. In conclusion, geometrical reconstruction by CTA yields similar results to ICA in terms of segment-based ESS calculation in patients with low and intermediate stenoses. Thus, it has the potential of allowing combined local hemodynamic and plaque morphologic information for risk stratification in patients with coronary artery disease.