Objective To investigate of effects of dipeptidyl peptidase- Ⅳ （DPP-4） inhibitor saxagliptin on cognitive function in rats with diabetes mellitus complicated with depression（ DD）， and possible mechanisms. Methods A total of 48 specific pathogen free（ SPF） healthy male SD rats were randomly divided into control group， model group， positive drug group and saxagliptin group， with 12 rats in each group. DD model was established by high-fat emulsion by gavage for 14 days， tail vein injection of streptozotocin（ STZ）， and chronic stress for 28 days. In the same period of chronic stress， the rats in the positive drug group were given metformin and fluoxetine capsules by gavage， in the saxagliptin group were given saxagliptin by gavage， and in the model group and the control group were given the same amount of normal saline， once a day for 28 days. The open field experiment and the Morris water maze test were used to detect the ability of exploring the unfamiliar environment and the ability of learning and memory. Morphological changes of hippocampus in rats were observed by hematoxylin-eosin（ HE） staining. The neuronal apoptosis of hippocampus was detected by TdT mediated dUTP nick end labeling（ TUNEL） staining. The levels of IL-1β， IL-6 and TNF-α in hippocampus tissues were detected by enzyme-linked immunosorbent assay（ ELISA）. Rusults Compared with the control group， the number of horizontal crossing grids of the rats in the model group， in the positive drug group and in the saxagliptin group［ （62.25±5.53），（ 85.92±4.58） and（ 78.25±6.02） vs.（ 104.75±6.11）］， and the number of standing vertically［ （7.33±1.97），（ 11.17±1.34） and（ 9.33±2.27） vs.（ 17.50±1.68）］ were decreased（ all P ＜ 0.05）. The escape latency was prolonged［ （51.75±8.01）s，（ 29.58±5.14）s and （35.42±3.48）s vs.（ 18.17±3.43）s； all P ＜ 0.01］， while the target quadrant memory time was shortened ［（51.75±8.01）s，（ 29.58±5.14）s and（ 35.42±3.48）s vs（. 18.17±3.43）s； all P ＜ 0.01］. The neuronal apoptosis index in hippocampus tissues were increased［ （74.66±4.62）%，（ 47.23±6.24）% and（ 39.22±2.80）% vs.（ 9.22±0.87）%； all P ＜ 0.01］， and the levels of IL-1β［ （41.26±5.82）ng/g，（ 21.01±3.84）ng/g and （29.89±7.42）ng/g vs.（ 12.22±5.02）ng/g］， IL-6［ （80.792±3.59）ng/g，（ 50.73±7.18） ng/g and（ 61.69± 6.29）ng/g vs.（ 31.90±4.79）ng/g］and TNF-α［ （90.94±8.72）ng/g，（ 66.57±9.93）ng/g and（ 77.70±6.96）ng/g vs. （49.88±5.24）ng/g］in the hippocampus tissues were increased（ P＜ 0.05 or P＜ 0.01）. Compared with the model group， in the positive drug group and the saxagliptin group， the number of horizontal crossing grids and number of standing vertically were increased（ all P＜0.01）. The escape latency was shortened（ all P＜0.01）， while the target quadrant memory time was prolonged（ all P＜ 0.01）. The neuronal apoptosis index in the hippocampus tissues was decreased（ all P＜ 0.01）， and the levels of IL-1β， IL-6 and TNF-α in the hippocampus tissues were decreased（ P ＜ 0.01）. Conclusions DPP-4 inhibitor Saxagliptin can improve cognitive impairment in rats with DD， and its mechanism may be related to the reduction of hippocampal damage by reducing the inflammation of hippocampus.