Background: There is limited data to determine accuracy and precision of detecting acute coronary thrombus via angiography. Objective: To build a parsimonious model for the prediction of acute coronary thrombus from four commonly used angiographic characteristics (spherical, ovoid, or irregular filling defect; abrupt vessel cutoff with persistence of contrast; intraluminal staining; any coronary filling defect). Methods: Angiographic Core Laboratory (Baltimore, Maryland) evaluated angiograms of 80 acute MI / stable CAD subjects in blinded fashion. Outcome was defined as presence of thrombus by histological examination. Multivariable generalized linear modeling was conducted to determine best combinations of characteristics for discriminating between lesions with a histologically confirmed thrombus versus lesions without a histologically confirmed thrombus. Results: A best-fit parsimonious model revealed that individual presence of abrupt vessel cutoff with persistence of contrast (O.R=16, P<0.0001) or any coronary filling defect (O.R=84, P<0.0001), or both in concurrence (O.R=105, P<0.0001) (Figure 1 and Table 1), was most significantly associated with the presence of coronary thrombus. Conclusions: Modeling suggests that the presence of abrupt vessel cutoff with persistence of contrast or any filling defect, individually or in concurrence, is significantly associated with acute coronary thrombus, and therefore warrants validation in an independent cohort.