Abstract Background The GRACE risk score is generically recommended by guidelines for timing of invasive coronary angiography without stating which score should be used. The aim was to determine the diagnostic performance of different GRACE risk scores in comparison to the ESC 0/1 h-algorithm using high-sensitivity cardiac troponin (hs-cTn). Methods Prospectively enrolled patients presenting with symptoms suggestive of myocardial infarction (MI) in two large studies testing biomarker diagnostic strategies were included. Five GRACE risk scores were calculated. The amount of risk reclassification and the theoretical impact on guideline-recommended timing of invasive coronary angiography was studied. Results Overall, 8,618 patients were eligible for analyses. Comparing different GRACE risk scores, up to 63.8% of participants were reclassified into a different risk category. The proportion of MIs identified (i.e., sensitivity) dramatically differed between GRACE risk scores (range 23.8–66.5%) and was lower for any score than for the ESC 0/1 h-algorithm (78.1%). Supplementing the ESC 0/1 h-algorithm with a GRACE risk score slightly increased sensitivity (P < 0.001 for all scores). However, this increased the number of false positive results. Conclusion The substantial amount of risk reclassification causes clinically meaningful differences in the proportion of patients meeting the recommended threshold for pursuing early invasive strategy according to the different GRACE scores. The single best test to detect MIs is the ESC 0/1 h-algorithm. Combining GRACE risk scoring with hs-cTn testing slightly increases the detection of MIs but also increases the number of patients with false positive results who would undergo potential unnecessarily early invasive coronary angiography. Graphical abstract