Methods Totally, 24 BALB/c mice were assigned to the AR group, control group, GzmB group, and blank group (each n = 6). The blank group was normally fed without treatment, and the other three groups were treated by ovalbumin (OVA) to induce AR models, in which the GzmB group was intranasally injected with lentiviral vector suppressing GzmB expression during the second immunization, while the control group was given the GzmB-blank vector. The times of AR pathological behaviours such as sneezing and scratching the nose of mice were observed and counted. The nasal lavage fluid of each mouse was acquired, and then, the mouse was executed by cervical dislocation, followed by collection of blood and nasal mucosa tissues. Then, ELISA was adopted for quantifying immunoglobulin E (IgE), interleukin (IL)-4, IL-6, and histamine (HA), and nasal mucosa tissues were treated by HE and TUNEL staining to observing their histopathological manifestations. PCR and western blot (WB) were adopted for quantifying GzmB and miR-378a-3p. Additionally, with NP69 cells, dual luciferase reporter (DLR) assay was carried out for determining the targeting association of GzmB with miR-378a-3p. Another 24 mice were assigned to the AR group, GzmB group, miR-378a-3p group, and GzmB+ miR-378a-3p group (each n = 6). The AR and GzmB groups were treated as above. The miR-378a-3p group was intervened by lentiviral vector suppressing miR-378a-3p, while the GzmB+ miR-378a-3p group was given GzmB and lentiviral vector suppressing miR-378a-3p meantime. A rescue assay was conducted through repeating the above tests. Results The times of sneezing and rubbing the nose and the levels of IgE, IL-4, IL-6, and HA were similar between the control and AR groups (all P > 0.05), and these items of the two groups were all higher than those of the blank and GzmB groups (all P < 0.05). However, no notable difference was observed in IL-4 and IL-6 levels between the GzmB and blank groups (both P > 0.05), while higher levels of other detection results were found in the former group than in the latter (all P < 0.05). The staining results revealed obvious congestion, oedema, and necrosis structures in the nasal mucosa epithelium of the control and AR groups and also revealed a large number of infiltrating eosinophils and notable increase of apoptotic nasal mucosa epithelial cells. The GzmB group showed notably improved nasal mucosa tissues, and its infiltration and apoptosis of eosinophils were more notable than those of the blank group, but notably weaker than those of the AR and control groups. Additionally, the PCR and WB results revealed similar miR-378a-3p and GzmB levels in nasal mucosa between the control and AR groups (both P > 0.05), and a notable decrease of miR-378a-3p and a notable increase of GzmB in both groups (both P < 0.05). The DLR assay revealed notably suppressed fluorescence activity of GzmB-WT in NP69 cells after transfection of miR-378a-3p mimics (P < 0.05) and notably down regulated GzmB protein after increase of miR-378a-3p (P<0.05). Finally, the rescue assay revealed that downregulating miR-378a-3p aggravated the pathological changes of AR (P < 0.05) and also completely reversed the impacts of inhibiting GzmB on the pathological behaviours of AR mice. Conclusions MiR-378a-3p can accelerate the pathological development of AR through targeted inhibition on the release of pro-inflammatory factors such as IgE and HA activated by GzmB, so it is a promising molecular target of AR therapy and offers a novel research direction for the complete cure of AR.