SARS-CoV-2 induces a large range of neurological symptoms even without respiratory manifestations. However, there is still no consensus concerning viral entry and viral tropism in the brain as well as viral-induced inflammation. Here, we used a well-characterized culture of human neuronal/glial cells (hNGCs) differentiated from fetal neural progenitors to provide new arguments about SARS-CoV-2 infection in human brain. We showed that astrocytes were highly permissive to the virus, confirming Andrews et al., (2022) who used human cortical organotypic slices and brain organoids. In hNGCs, viral infection led to a strong alteration of neuronal morphology and to astrocytes death. SARS-CoV-2 infected-hNGCs were then used to screen twenty molecules (viral polymerase inhibitors, statins, antimalaria or antiparasitic drugs, etc…) with either already known or unknown antiviral activity against this virus. Image analyses, quantification of viral RNA and viral particles were used to determine the anti-viral efficiency. We revealed that seven molecules with known anti-SARS-CoV-2 activity in other cell types were inefficient in hNGCs whereas the others were confirmed, which demonstrated that the efficiency of these molecules was cell-type dependent. Among previously unknown antiviral molecules against SARS-CoV-2, we found that small-molecule cyclophilin inhibitors have an antiviral activity in infected-hNGCs. Our data thus demonstrate that human astrocytes are permissive to the virus and that infection strongly affects human neuronal/glial cells. They also outline the importance of using cellular model that are brain specific to question the role of antiviral molecules in this organ.