Late gadolinium enhancement (LGE) is widely used to precisely localize and determine the extent and transmurality of myocardial scarring. Performing LGE imaging at end-systole may reduce motion artefacts. It is therefore important to know whether end-systolic imaging influences infarct transmurality in patients with ischemic scar. 107 dysfunctional segments were studied in 20 consecutive patients with established coronary artery disease. Patient specific trigger delays were used to obtain end-diastolic and end-systolic LGE images (LGE(ed), LGE(es)). Wall thickness (WT(ed), WT(es)), thickness of the remaining viable rim (RIM(ed), RIM(es)) and end-diastolic scar thickness were measured manually. There was LGE in 84% of all dysfunctional segments with a mean scar of 3.4 ± 2.5 mm. Total wall thickness and the thickness of the remaining viable rim increased from diastole to systole (WT(ed) 7.8 ± 1.9 vs. WT(es) 8.4 ± 2.2; P < 0.001; RIM(ed) 4.4 ± 3.1 vs. RIM(es) 5 ± 3.4; P < 0.001). Transmurality of scar decreased from end-diastole to end-systole (LGE(ed) 46 ± 33% vs. LGE(es) 44 ± 33%; P < 0.001). This was most pronounced in a subgroup of segments (n = 15) with visual scar transmurality between 50 and 75% (LGE(ed) 75 ± 15% vs. LGE(es) 70 ± 16%; P < 0.001). The change in transmurality was inversely correlated with the change of the thickness of the remaining viable rim between diastole and systole (r = -0.7; P < 0.001). Scar transmurality was reduced by up to 12% in the individual patient. Scar transmurality changes due to thickening of the remaining viable rim. Whereas these differences might not impact on clinical decision-making in most patients, there will be an occasional misclassification if cut-off values are used (e.g. scar <50 or >50%) or if exact data is required for research studies.