Competing definitions and classifications of coronary artery abnormalities (CAAs) after Kawasaki disease (KD) have been arbitrarily defined based on clinical experience. We sought to propose a classification system for CAAs based only on coronary artery z-scores. All echocardiograms performed between 1990 and 2007 on patients with a previous history of KD were reviewed. Coronary artery luminal dimensions were converted to body-surface-area-adjusted z-scores and compared to current classification systems. A total of 1356 patients with a previous history of KD underwent 4379 echocardiograms. There was important overlap in the distributions of coronary artery z-scores between the different CAA classes as defined by the American Heart Association (AHA). The AHA classification underestimated the severity of CAAs in 19–32% of small CAAs and 35–78% of medium CAAs. We determined the optimal definition of CAA to be small if the z-score is ≥2.5 to <5.0, large if the z-score is ≥5.0 to <10.0, and giant if the z-score is ≥10.0. This classification seems to appropriately apply to the circumflex branch despite a lack of normal values for this branch. The current AHA classification might not accurately classify CAAs in KD patients. Accurate classification is important for defining management and prognosis consistently across patient age and size.