Abstract Background After rule-out of MI in ED chest pain patients, uncertainty exists with respect to which remaining patients would benefit from objective functional/anatomical testing to detect potential underlying CAD. Using proteomics, we experimentally identified IGFBP-3 as a biomarker to assist this risk stratification and applied its measurement to >3,000 ED patients from 3 separate studies. Methods An isolated rat heart perfusate model coupled to mass spectrometry identified markers of cardiac ischemia. Human cardiac ischemia was carried out using cardiac exercise stress testing (EST, n=12). Temporal changes in IGFBP-3 during experimental MI were determined in patients undergoing septal alcohol ablation (SAA, n=12) and arterio-venous gradients of IGFBP-3 were determined from cases of proscribed cardiac catheterisation (CC, n=14). The ability of IGFBP-3 to assist clinical assessment (ROC analyses, logistic regression, C-statistic, pathways analyses) and ED risk assessment by the No Objective Testing (NOT) rule was carried out on samples from 3 prospectively recruited studies (SPACE, NZ, n=766; APACE, CH, n=1,056; FAST-TRAC, USA, n=1,230). Results IGFBP-3 identified by MS was present in perfusates of ischemic hearts only (n=3). In human cardiac EST, a delta of IGFBP-3 (∆IGFBP-3) between 0-150minutes was seen in positive ischemic tests, but not in controls (n=6 per group, p=0.026). In SAA patients, peripheral IGFBP-3 levels did not change over 24 hours (n=12, p=0.573) whereas in CC patients, significant extraction gradients of IGFBP-3 were seen across the heart and liver (p=0.013, n=14). In ED chest pain patients identified by the NOT rule (n=1,708), ΔIGFBP-3 between presentation and +2 hours (∆ IGFBP-3) was an independent logistic model predictor of unstable angina (n=123) alongside age, sex, cholesterol and Hx of CAD/MI. At 90% specificity, ∆IGFBP-3 improved the sensitivity of the model from 31% to 43%, the PPV 20% to 25% and the NPV from 94% to 95% (p=0.020). IGFBP-3 made a small improvement in "low risk" NOT rule patient identification (29% to 30%), but made a larger improvement to the ROC curve based identification in NOT rule patients for the finding of stenosis >70% on angiography (n=101, AUC clinical model = 0.682; model plus ∆ IGFBP-3 = 0.725, p=0.044). Pathways analyses (that included hs-TnT) in angiography patients (n=224) revealed ∆IGFBP-3 generated a 28% relative increase in low risk case identification (n=46 to 59). Conclusions IGFBP-3 is identified as a novel potential biomarker of cardiac ischemia, that is taken up/degraded across the heart and appears less responsive to complete infarction (MI). 0-2 hour changes in IGFBP-3 (∆ IGFBP-3) in patients identified as eligible for ED based NOT rule inclusion show promise as a useful tool to identify those with significant CAD, evidenced by the diagnosis of unstable angina, positive EST results or findings of 70% occlusion on angiography.