To investigate the protective effect and mechanism of total flavonoids from Citrus paradise cv. Changshan-huyou extracts (TFC) on oxygen-glucose deprivation (OGD) of primary neurons and chronic ischemia-induced cerebral injury in mice. Primary hippocampal neurons of 18-day fetal rats were isolated and cultured for 1 week, then treated with 0.25, 0.50 and 1.00 mg/mL TFC. After oxygen-glucose deprivation for 1 h, cells were reperfused for 6 and 24 h, respectively. The cytoskeleton was observed by phalloidin staining. In animal study, 6-week ICR male mice were randomly divided into sham operation group, model group, low-dose (10 mg/kg), medium-dose (25 mg/kg) and high-dose (50 mg/kg) TFC treatment groups, with 20 mice in each group. After 3 weeks, chronic cerebral ischemia was induced by unilateral common carotid artery ligation in all groups except sham operation group. Mice were treated with different concentrations of TFC in the three TFC treatment groups for 4 weeks. Open field test, novel object recognition test and Morris water maze test were used to evaluate anxiety, learning and memory of these mice. Nissl, HE and Golgi stainings were used to detect neuronal degeneration and dendritic spine changes in the cortex and the hippocampus. The expression levels of Rho-associated kinase (ROCK) 2, LIM kinase (LIMK) 1, cofilin and its phosphorylation, as well as the expression of globular actin (G-actin) and filamentous actin (F-actin) protein in hippocampus of mice were detected by Western blotting. Neurons subjected to OGD showed that neurites displayed shortening and breakage; while treatment with TFC reversed OGD-induced neurite injury, especially in the 0.50 mg/mL TFC group. Compared with the sham operation group, the mice in the model group showed a significant decline in anxiety and cognitive ability (P<0.01), whereas treatment with TFC significantly reversed anxiety and cognitive deficits (P<0.05). Improvement in the medium-dose TFC group was the most obvious. Histopathological analysis indicated that the number of Nissl bodies and dendritic spines in hippocampus and cortex were decreased in the model group (all P<0.01). However, after treatment with medium dose of TFC, the number of Nissl bodies and dendritic spines (all P<0.05) was significantly recovered. Compared with the sham operation group, the phosphorylation level of ROCK2 in the brain tissue of the model group was significantly increased (P<0.05), while the phosphorylation levels of LIMK1 and cofilin were significantly decreased (P<0.05), and the relative content ratio of G-actin/F-actin was significantly increased (P<0.05). After administration of TFC, the phosphorylation level of ROCK2 in brain tissue of each group was significantly decreased (P<0.05), while the phosphorylation levels of LIMK1 and cofilin were significantly up-regulated (P<0.05) and the relative content ratio of G-actin/F-actin was significantly decreased (P<0.05). TFC protects from ischemia-induced cytoskeletal damage, reduces neuronal dendritic spine injury and protects mice against chronic cerebral ischemia through RhoA-ROCK2 signaling pathway, indicating that TFC might be a potential candidate for treatment of chronic ischemic cerebral injury.