State-of-the-art virus detection technology has advanced a lot, yet technology to evaluate the impacts of viruses on bee physiology and health is basically lacking. However, such technology is sorely needed to understand how multi-host viruses can impact the composition of the bee community. Here, we evaluated the potential of hemolymph metabolites as biomarkers to identify the viral infection status in bees. A metabolomics strategy based on ultra-high-performance liquid chromatography coupled to high-resolution mass spectrometry was implemented. First, we constructed a predictive model for standardized bumble bees, in which non-infected bees were metabolically differentiated from an overt Israeli acute paralysis virus (IAPV) infection (R2Y = 0.993; Q2 = 0.906), as well as a covert slow bee paralysis virus (SBPV) infection (R2Y = 0.999; Q2 = 0.875). Second, two sets of potential biomarkers were identified, being descriptors for the metabolomic changes in the bee’s hemolymph following viral infection. Third, the biomarker sets were evaluated in a new dataset only containing wild bees and successfully discriminated virus infection versus non-virus infection with an AUC of 0.985. We concluded that screening hemolymph metabolite markers can underpin physiological changes linked to virus infection dynamics, opening promising avenues to identify, monitor, and predict the effects of virus infection in a bee community within a specific environment.