We sought to elucidate the risk profiles of patients with Kawasaki disease who developed coronary artery abnormalities through a retrospective analysis with special reference to steroid treatment. Demographics of the patients were obtained from medical records, and characteristics of the coronary artery abnormalities were evaluated by echocardiography and coronary angiography, which included number, location, size, and length of coronary artery abnormalities (we evaluated by cardiac catheterisation with the American Heart Association classification with segments). We divided the patients into two groups based on steroid use and compared their characteristics and the complications of coronary artery abnormalities and cardiac events. A total of 29 patients were diagnosed with coronary artery abnormalities by echocardiography and coronary angiography during the study period (24 male; median age, 24 months [range: 2–84 months]). Eighteen patients were treated with aspirin and intravenous immunoglobulin (63%, non-steroid group), whereas 11 received aspirin and intravenous immunoglobulin plus steroids (37%, steroid group). No significant differences were found in the number and location of coronary artery abnormalities between the steroid and non-steroid groups. However, the size and number of segments for coronary artery abnormalities were significantly larger and shorter, respectively, in the steroid group (z-score: non-steroid group 6.3 versus steroid group 8.7; p < 0.01). The coronary artery abnormality segments under steroid use were also shorter (non-steroid group versus steroid group, two segments versus one segment; p = 0.02). Coronary artery abnormality size was larger in patients who used steroids than that of non-steroids. This study showed that steroid use significantly affected coronary artery abnormality size in patients with Kawasaki disease. However, cardiac complications from coronary artery abnormalities and cardiac events were comparable between the steroid and non-steroid groups. Further prospective, multicentre studies are needed to confirm these findings.