Delayed neurocognitive recovery (dNCR), a postoperative complication that occurs in elderly patients, still lacks effective treatment. Interleukin-33 (IL-33) has been proved to modulate neuroinflammation and synaptic plasticity, among other effects, but the role of IL-33 in dNCR is not clear. We established a dNCR model in aged mice by laparotomy under sevoflurane anesthesia. Cognition was evaluated by Morris water maze (MWM) and fear conditioning test (FCT). Immunofluorescence was used to detect the density of IL-33 and glial fibrillary acidic protein (GFAP) co-localization, ionized calcium-binding adapter molecule 1, vesicular glutamate transporter 1 (vGlut1) and postsynaptic density protein-95 (PSD95) co-localization in the hippocampus. IL-33, GFAP, vGlut1 and PSD95 were tested by Western blotting. Enzyme-linked immunosorbent assay was used to detect the levels of tumor necrosis factor-alpha (TNF-α), interleukin-1 beta (IL-1β) and IL-10. Surgery/anesthesia reduced the level of IL-33 in the hippocampus. Intraperitoneal injection of 200 ng IL-33 per mouse significantly decreased the latency to the platform and increased the number of platform crossings and the target quadrant dwell time in MWM, while increasing the freezing time in the context test of FCT. Furthermore, IL-33 inhibited microglial activation and the release of TNF-α and IL-1β while upregulating the markers of excitatory synapses vGlut1 and PSD95. Our findings indicated that IL-33 improved cognition by inhibiting the hippocampal inflammatory response and upregulating the number of excitatory synapses. Therefore, IL-33 is a potential drug for the treatment of dNCR.