Abstract: Objective To evaluate the predictive value for serum albumin-globulin ratio （A/G） for the risk of Kawasaki disease （KD） in febrile children. Methods In this single-center retrospective cohort study, the big-data platform of a certain third-tier A-level hospital was employed. Clinical and laboratory data of 432 febrile children aged 0-16 were collected. The children were divided into 4 groups based on the quartiles and median of their serum A/G ratios. The correlation between the A/G and the risk of KD was analyzed. Results The total incidence rate of KD in 432 children with fever was 14.4% （62/432）.A significant negative correlation was observed between the A/G and the risk of KD. For each increase of 0.1 in the A/G, the risk of KD was decreased by 18% （OR=0.82, 95% CI 0.76-0.88, P < 0.001）. Subgroup analysis showed a consistent negative correlation between the A/G and KD risk across different ages, genders, and biochemical markers. Conclusions A/G may be a useful biomarker for assessing the risk of KD in febrile children. As the A/G increases, the risk of KD is gradually decreased. This finding has potential clinical application value for early diagnosis and risk assessment of KD.