There are significant controversies on the antibacterial properties of graphene oxide (GO): GO was reported to be bactericidal in saline, whereas its activity in nutrient broth was controversial. To unveil the mechanisms underlying these contradictions, we performed antibacterial assays under comparable conditions. In saline, bare GO sheets were intrinsically bactericidal, yielding a bacterial survival percentage of <1% at 200 μg/mL. Supplementing saline with ≤10% Luria-Bertani (LB) broth, however, progressively deactivated its bactericidal activity depending on LB-supplementation ratio. Supplementation of 10% LB made GO completely inactive; instead, ∼100-fold bacterial growth was observed. Atomic force microscopy images showed that certain LB components were adsorbed on GO basal planes. Using bovine serum albumin and tryptophan as well-defined model adsorbates, we found that noncovalent adsorption on GO basal planes may account for the deactivation of GO's bactericidal activity. Moreover, this deactivation mechanism was shown to be extrapolatable to GO's cytotoxicity against mammalian cells. Taken together, our observations suggest that bare GO intrinsically kills both bacteria and mammalian cells and noncovalent adsorption on its basal planes may be a global deactivation mechanism for GO's cytotoxicity.