Background <p>The metabolic score for insulin resistance (METS-IR) is considered a novel, non-invasive, and validated assessment of insulin resistance (IR), which may contribute to a poor neurological prognosis after stroke. However, evidence on the correlation between METS-IR and stroke prognosis in populations with acute ischemic stroke (AIS) remains limited. Therefore, this research aims to assess the effect of METS-IR levels on three-month unfavorable outcomes after AIS.</p> Methods <p>This prospective cohort study enrolled 1,722 AIS subjects who received treatment at a Korean hospital. The effect of METS-IR levels on three-month unfavorable outcomes was analyzed using a binary logistic regression model, while smoothed curve fitting and generalized additive modeling were applied to investigate non-linear associations.</p> Results <p>There was a non-linear correlation between METS-IR and three-month unfavorable outcomes after AIS was identified in this study. An inflection point in this relationship was observed at a METS-IR value of 35.16. For METS-IR values ≤ 35.16, a significant negative correlation was detected, with higher METS-IR levels being associated with a reduced risk of three-month unfavorable outcomes (OR: 0.93, 95% CI: 0.89–0.97, <i>P</i> = 0.0009). Conversely, for METS-IR values above this threshold, the association was no longer statistically significant (OR: 1.02, 95% CI: 0.99–1.05, <i>P</i> = 0.2486).</p> Conclusion <p>Our findings revealed a non-linear negative correlation between the effect of METS-IR levels on three-month unfavorable prognosis in AIS subjects.</p>