Background: Evidence supports the role of host genetic diversity for the clinical course variation of coronavirus disease 2019 (COVID-19). Variation in the cannabinoid CB2 receptor gene (CNR2) could affect the endocannabinoids regulatory actions on the immune system, resulting in an increased risk of various inflammatory diseases. The present study investigated the relationship between the CNR2 rs35761398 (Q63R) functional variation and COVID-19 severity. Results: A total of 200 Iranian COVID-19 patients (100 expired and 100 discharged) were enrolled in the study and genotyped through TaqMan assay. The co-dominant, dominant, recessive, over-dominant, and additive inheritance models were analyzed using SNPStats software. In silico molecular docking was also performed to simulate the effects of Q63R variation on CB2 binding with a ligand and with G-protein. A significant difference in the Q63R allele and genotype distributions was found between COVID-19 expired and discharged patients in co-dominant (OR: 3.33, 95% CI: 1.25-8.88, p = 0.043), recessive (OR: 2.92, 95% CI: 1.16-7.33, p = 0.017), and additive inheritance (OR: 1.62, 95% CI: 1.06-2.48, p = 0.025) models. The molecular docking results showed that the predicted structure of mutant CB2 (63R type) could not bind to G-protein in the correct position.Conclusions: The data implied the involvement of the CNR2 gene in the severity of COVID-19 in Iranian patients. Identification of genes related to susceptibility and severity of COVID-19 may lead to specific targets for repurposing or drug development.