Objective To investigate the effect of dexmedetomidine (Dex) on delayed neurocognitive recovery (DNR) in elderly patients after surgery and related mechanism. Methods A total of 186 elderly patients who underwent non‑cardiac, non‑neural, and non‑transplantation operations under general anesthesia were enrolled. According to the random number table method, they were divided into two groups: a dexmedetomidine group (group Dex, n=94) and a control group (group C, n=92). According to the condition of patients, group Dex was further divided into two groups: a Dex‑maintenance dose group (n=48, with sinus bradycardia, sick sinus syndrome, conduction block, or ventricular insufficiency before operation) and a Dex‑loading dose group (n=46, without the above diseases). Group C was further divided into two groups: a C‑maintenance dose group (n=47, with sinus bradycardia, sick sinus syndrome, conduction block or ventricular insufficiency before operation) and a C‑loading dose group (n=45, without the above diseases). In addition to the same basic anesthetics in each group, the Dex‑maintenance dose group was treated with a maintenance dose of Dex at 0.4 μg·kg-1·h-1, while the Dex‑loading dose group was pumped with a loading dose of Dex at 1.0 μg/kg within 10 min before surgery, followed by a maintenance dose of Dex at 0.4 μg·kg-1·h-1. The C‑loading dose group was administered with the same dose of normal saline as the Dex‑loading dose group, while the C‑maintenance dose group was administered with the same dose of normal saline as the Dex‑maintenance dose group. The preoperative, intraoperative and postoperative information of patients was recorded. Their Montreal Cognitive Assessment (MoCA) scores were evaluated three days before operation (T1), and at postoperative 1 d (T2), 2 d (T3), and 3 d (T4). The incidence of DNR within 3 d and 30 d after surgery was recorded. The concentrations of glial fibrillary acidic protein (GFAP), neuron specific enolase (NSE), IL‑6, and phosphorylated tau protein were recorded at T1 and T3, and the increases were calculated (from T3 to T1). The adverse events within 30 d after operation was also recorded. The dose and dosing of Dex used by all patients during operation were collected and the Tivatrainer software was used to generate a blood dose‑time curve of Dex. Results Group Dex showed a higher incidence of diabetes than group C (P<0.05), and the Dex‑maintenance dose group presented a higher incidence of diabetes than the Dex‑loading dose group (P<0.05). There was no statistical difference in intraoperative and postoperative information between group Dex and group C, and between the Dex‑loading dose group and the Dex‑maintenance dose group (P>0.05). The incidence of DNR in group Dex within 3 d and 30 d after operation was lower than that in group C (P<0.05). The incidence of DNR in the Dex‑loading dose group within 3 d after operation was lower than that in the Dex‑maintenance dose group (P<0.05), but there was no statistical difference in the incidence of DNR within 30 d between the two groups (P>0.05). Within 30 d after operation, one case of pulmonary complications occurred in both group C and group Dex. Two patients in group C and one patient in group Dex underwent secondary surgery, without statistical difference in adverse events between the two groups (P>0.05). The increases of NSE and IL‑6 in patients with DNR after operation were more than those in patients without DNR (P<0.05), while the increases of NSE and IL‑6 in group Dex were less than those in group C (P<0.05). The Dex‑loading dose group reached the effective blood concentration of Dex at 0.153 μg/L faster than the Dex‑maintenance dose group, resulting in a longer duration of maintenance. Conclusions Dex can reduce the incidence of DNR in elderly patients within 3 d after operation, which may be related to inhibiting the increase of IL‑6 and NSE after operation and reducing neuroinflammation. Treatment with Dex at a loading dose of 1.0 μg/kg can reach the effective blood concentration faster and maintain for a long time, in order to exert more significant brain protective effect than the maintenance dose.