OBJECTIVE To explore the relationship between the concentration of serum magnesium and the short-term outcome of patients with acute ischemic stroke, in order to provide evidence for improving the outcomes. METHODS Patients with acute ischemic stroke under study, were from four hospitals in Shandong province. Data on demographic characteristics, life style related risk factors, history of cardiovascular disease, blood pressure at admission and other clinical characteristics were collected for all the participants. The outcomes were defined as National Institutes of Health Stroke Scale (NIHSS) ≥ 10 or death. According to NIHSS, the subjects were divided into two groups: death/NIHSS ≥ 10 and NIHSS < 10. Concentrations of Mg(2+) were categorized into four levels according to the quartiles of serum magnesium. Cox proportion hazard regression analysis was used to evaluate the association between serum magnesium concentrations and the short-term outcome of acute ischemic stroke. RESULTS In the death/NIHSS ≥ 10 group, concentrations of serum magnesium and the time from onset to admission were lower than that in the NIHSS < 10 group while the systolic blood pressure on admission, the proportion of low density lipoprotein abnormal, impaired fasting glucose and history of auricular fibrillation were all higher than that in the NIHSS < 10 group. Without the adjustment of multiple factors, when comparing to the lowest quartile of serum magnesium level, the fourth quartile (highest) seemed to have had a tendency of reducing the risk of death/NIHSS ≥ 10 (RR = 0.47, P < 0.05). When multiple factors were adjusted (adjust serum calcium, potassium level and other factors), the fourth and the third quartiles could both reduce the risk of death/NIHSS ≥ 10 (RR values were 0.39 and 0.54, P < 0.05, respectively). With or without the adjustment of multiple factors, there appeared a dose-response relationship between serum magnesium concentrations and the risk to death/NIHSS ≥ 10 (trend P < 0.05). CONCLUSIONS Higher serum magnesium concentrations could reduce the risk to death/NIHSS ≥ 10, suggesting that there was a dose-response relationship between magnesium and the risk to death/NIHSS ≥ 10.