Objective To investigate the effect of Salvianolic acid on endoplasmic reticulum stress (ERS) pathway in brain hippocampus of PAP mice. Methods Twenty PAP dual transgenic male mice were selected, they were randomly divided into a PAP mice model group and a Salvianolic acid group, 10 mice in each group; another 10 SPF grade C57BL/6J male mice were selected as a normal control group. In the Salvianolic acid group, 0.9% normal saline solution of Salvianolic lyophilized injection (400 g/L) of dosage 21 mg·kg-1·d-1 was injected intravenously through a tail vein of mice; the PAP mice model and normal control groups were given the same amount of 0.9% normal saline, and the therapeutic course was consecutive 4 weeks in the three groups. At the end of the 4th week, the Morris water maze test was carried out to observe the changes of escape latency, the third quadrant residence time (RTQ), entry angle into water and cross-platform times of mice in each group; amyloid precursor protein (APP) positive cell expression in cerebral hippocampus of mice were detected by immunohistochemistry; Western Blot was used to detect the expression level of PER like endoplasmic reticulum kinase-eukaryon initiation factor 2α-C/EBP homogenous protein (PERK-eIF2α-CHOP) pathway related proteins in hippocampus of mice. Results The escape latency of the PAP mice model group on the 1st to 5th day were significantly longer than those of the normal control group, although a downward trend was observed on the 5th day, it was still significantly longer than that of the model group (seconds: 58.41±2.36 vs. 28.60±10.15); compared with the PAP mice model group, the escape latency of Salvianolic acid group was shorter at each time point, and reached the shortest level on the 5th day (seconds: 31.97±8.36 vs. 58.41±2.36). In the PAP mice model group, the RTQ and the number of crossing platforms were significantly lower than those in the normal control group [RTQ (seconds): 8.27±2.95 vs. 15.97±7.33, numbers of crossing platforms (frequency/90 s): 0.70±0.95 vs. 2.70±0.48]; the entry angle was obviously greater than that of the normal control group [(47.94±4.68)°vs. (32.66±2.55)°, P < 0.05]. Compared with PAP mice model group, in Salvianolic acid group, the RTQ and number of crossing platform were significantly higher [RTQ (seconds): 13.57±1.86 vs. 8.27±2.95, number of crossing platforms (frequency/90 s):1.60±0.97 vs. 0.70±0.47], the entry angle was markedly smaller [(35.46±6.79)°vs. (47.94±4.68)°,P < 0.05]. The positive expression rate of APP and the protein expressions of CHOP, p-eIF2α in PAP mice model group were significantly higher than those in the normal control group [the positive rate of APP: (60.44±6.19)% vs. (21.05±5.87)%, CHOP protein expression (gray value): 3.09±0.07 vs. 1.46±0.09, p-eIF2αprotein expression (gray value): 0.98±0.09 vs. 0.47±0.06, all P < 0.01], the expression of PERK and p-PERK were lower than those in normal control group [PERK (gray value): 0.42±0.06 vs. 0.82±0.11, p-PERK protein expression (gray value): 0.98±0.09 vs. 0.64±0.10, both P < 0.01]; the positive expression rate of APP and protein expressions of CHOP, p-eIF2α in Salvianolic acid group were significantly lower than those in PAP mice model group [positive expression rate of APP: (33.09±10.33)% vs. (60.44±6.19)%, CHOP protein expression (gray value): 1.57±0.12 vs. 3.09±0.07, p-eIF2α protein expression (gray value): 0.80±0.07 vs. 0.98±0.09, all P < 0.01], while PERK and p-PERK expression were significantly higher than those in the model group [PERK (gray value): 0.89±0.12 vs. 0.42±0.06, p-PERK (gray value): 0.78±0.08 vs. 0.98±0.09, both P < 0.01]. Conclusion Salvianolic acid might work through the PERK-eIF2α-CHOP pathway to reduce the retention of APP in the hippocampus tissue of PAP dual-transgenic mice, thereby the learning ability of the mice is improved, and the progression of brain injury delayed.