Objective To investigate the correlation between stress hyperglycemia ratio (SHR) and 90-day clinical prognosis of patients with acute ischemic stroke (AIS). Methods The data of AIS patients who were admitted in Department of Neurology, Yangpu Hospital, Tongji University from May 2016 to October 2020 were retrospectively analyzed. The outcome was assessed by 90-day mRS and composite vascular events (recurrent stroke, myocardial infarction and vascular death). The favorable prognosis was defined as a mRS score of 0-2, and unfavorable prognosis was defined as a mRS score of 3-6 and composite vascular events within 90 days after stroke. SHR was calculated by fasting blood glucose divided by the estimated average blood glucose. The multivariate logistic regression analysis was used to determine whether SHR was an independent influencing factor of the 90-day prognosis. ROC curve was used to evaluate the effect of SHR to predict 90-day prognosis. Spearman correlation was used to analyze the correlation between SHR and clinical factors. Results (1) A total of 1484 patients were included,with the median age of 70 (62-80) years and 948 males (63.9%). The median baseline NIHSS score was 3 (2-7). There were 923 cases (62.2%) in favorable prognosis group and 561 cases (37.8%) in unfavorable prognosis group. (2) Univariate analysis showed that the SHR in favorable prognosis group was significantly lower than that in unfavorable prognosis group [0.82 (0.72-0.95) vs . 0.86 (0.74-1.02), P =0.001]. (3) Multivariate logistic regression analysis showed that compared with the lowest quartile of SHR, the highest quartile was independently associated with the 90-day unfavorable prognosis (OR 2.22, 95%CI 1.26-3.91, P =0.006), which had statistical difference in non-diabetic group in subgroup analysis (OR 2.11, 95%CI 1.19-3.75, P =0.010), and no statistical difference in diabetic group (OR 1.53, 95%CI 0.69-3.42, P =0.298). (4) The area under the ROC curve (C-value) of SHR for predicting 90-day unfavorable prognosis was 0.552 (95%CI 0.526-0.578, P =0.001, the optimal cut-off value was 0.84, the sensitivity was 52.94%, and the specificity was 56.12%); and in non-diabetic subgroup, the C-value was 0.600 (95%CI 0.563-0.637, P