Previous studies have shown a positive association between the arteriosclerosis index (AI) and future diabetes risk. However, evidence in non-obese populations is limited. This study investigates the relationship between AI and future diabetes risk in non-obese East Asian adults. This retrospective cohort study enrolled 95,402 non-obese adults from China and Japan. Participants had a mean age of 42.92 ± 12.24 years, with 51,295 (53.77%) being male. Median follow-up was 3.01 years. Cox proportional hazards models assessed the association between baseline AI and diabetes risk. Non-linear associations were explored using cubic splines and smoothed curves in Cox models. Sensitivity analyses were performed. After adjusting for covariates, a positive association was found between AI and diabetes risk in non-obese adults (HR 1.09, 95% CI 1.03-1.15, P = 0.0017). A non-linear relationship was identified, with an inflection point at 1.47. Below this point, HR was 5.87 (95% CI 1.20-28.63, P = 0.0287)
above, it was 1.07 (95% CI 1.02-1.13, P = 0.0115). Sensitivity analyses affirmed the robustness of these results. This study identifies a positive, non-linear association between the AI and diabetes risk in non-obese adults. Interventions targeting AI reduction could significantly lower the risk of future diabetes in this population.