Honey is recognized as an efficacious topical antimicrobial agent in the treatment of burns and wounds. The antimicrobial activity in some honeys depends on the endogenous hydrogen peroxide content. This study was aimed to determine whether honey's hydrogen peroxide level could serve as a honey-specific, activity-associated biomarker that would allow predicting and assessing the therapeutic effects of honey. Using a broth microdilution assay, I analyzed antibacterial activities of 42 Canadian honeys against two bacterial strains: Escherichia coli (ATCC 14948) and Bacillus subtilis (ATCC 6633). The MIC90 and MIC50 were established from the dose-response relationship between antibacterial activities and honey concentrations. The impact of H2O2 on antibacterial activity was determined (i) by measuring the levels of H2O2 before and after its removal by catalase and (ii) by correlating the results with levels of antibacterial activities. Canadian honeys demonstrated moderate to high antibacterial activity against both bacterial species. Both MIC90 and MIC50 revealed that the honeys exhibited a selective growth inhibitory activity against E. coli, and this activity was strongly influenced by endogenous H2O2 concentrations. Bacillus subtilis activity was marginally significantly correlated with H2O2 content. The removal of H2O2 by catalase reduced the honeys' antibacterial activity, but the enzyme was unable to completely decompose endogenous H2O2. The 25%-30% H2O2 "leftover" was significantly correlated with the honeys' residual antibacterial activity against E. coli. These data indicate that all Canadian honeys exhibited antibacterial activity, with higher selectivity against E. coli than B. subtilis, and that these antibacterial activities were correlated with hydrogen peroxide production in honeys. Hydrogen peroxide levels in honey, therefore, is a strong predictor of the honey's antibacterial activity.