Context Curcumin has a significant effect on cerebral ischaemia–reperfusion injury (CIRI). However, the underlying mechanism is less studied. Objective This study investigates the role and mechanism of curcumin in CIRI. Materials and methods CIRI model Sprague-Dawley rats were divided into model, positive control and curcumin low/middle/high dose (50, 100 and 200 mg/kg/d) groups ( n = 10 each). Drug intervention was administered by gavage once a day for 4 weeks. We calculated the neurobehavioural score and observed the cerebral infarct volume. Glial cytopathological changes were observed after haematoxylin–eosin staining. Apoptosis was detected by TUNEL (TdT mediated dUTP nick end labelling). Extracellular signal-regulated protein kinase ( ERK ), C/EBP-homologous protein ( CHOP ) and caspase-11 mRNA were detected by real-time PCR. Phosphorylated ERK (p-ERK), phosphorylated CHOP (p-CHOP) and caspase-11 were detected by Western blot. Superoxide dismutase (SOD) activity was detected by xanthine oxidation method; malondialdehyde (MDA) content by thiobarbituric acid colorimetry; and, glutathione (GSH) by spectrophotometry. Results Compared with control, the neurobehavioural scores, neuronal apoptosis, MDA, IL-1β, IL-18, mRNAs and protein levels of ERK/p-ERK, CHOP/p-CHOP and caspase-11 in model group were significantly higher ( p < 0.01). Compared with model, the positive control and medium/high dose curcumin groups were significantly lower ( p < 0.01). However, SOD and GSH decreased significantly in model group but increased significantly in positive control and medium/high dose curcumin groups ( p < 0.01). Moreover, curcumin significantly alleviated ischaemic state and neuroinflammation ( p < 0.01). Discussion and conclusions Curcumin may alleviate CIRI through ERK–CHOP–caspase-11 pathway. Our results may provide new insights into the pathogenesis of CIRI, and contribute to the development of treatment strategies for CIRI.