Viruses, including the recent, COVID-19-causing SARS-CoV-2 rely on their host for re-production. Here, we made use of genomic and structural information from SARS-CoV-2 and related viruses to create a biomass function capturing the stoichiometric amino and nucleic acid requirements of SARS-CoV-2. By incorporating this function into a stoichiometric metabolic model of the human cell and applying metabolic flux balance analysis, we identified host-based metabolic perturbations inhibiting SARS-CoV-2 reproduction. Our results highlight reactions in the central metabolism, as well as amino acid and nucleotide biosynthesis pathways. By incorporating host cellular maintenance into the model based on available protein expression data from human lung cells, we find that only few of these metabolic perturbations are able to selectively inhibit virus reproduction without significantly affecting host maintenance. Key reactions that are able to do so are found in metabolic junctions in amino acid biosynthesis pathways and in mitochondrial metabolite shuttles. We note that some of the catalysing enzymes of such reactions have demonstrated interactions with existing drugs, which can be used for experimental testing of the presented predictions. The developed computational approach offers a platform for rapid, experimentally testable generation of drug predictions against existing and emerging viruses based on their biomass requirements and their host’s metabolism.