Colorimetric assays have drawn increasing research interest with respect to the quantitative detection of hydrogen peroxide (H2O2) based on artificial enzymes because of their advantages with respect to natural enzymes, including design flexibility, low cost, and high stability. Regardless, the majority of the artificial enzymes exhibit low affinity to H2O2 with large Michaelis-Menten constants (Km). This indicates that the catalytic oxidation of 3,3',5,5'-tetramethylbenzidine (TMB) to blue-colored oxTMB requires a high H2O2 concentration, hindering the sensitivity of the colorimetric assay. To address this problem, novel reduced Co3O4 nanoparticles (R-Co3O4) have been synthesized in this study via a step-by-step procedure using ZIF-67 as the precursor. R-Co3O4 exhibits a considerably enhanced peroxidase-like activity when compared with that exhibited by pristine Co3O4 (P-Co3O4). The catalytic process in the case of R-Co3O4 occurs in accordance with the typical Michaelis-Menten equation, and the affinity of R-Co3O4 to H2O2 is apparently higher than that of P-Co3O4. Furthermore, the density functional theory calculations revealed that the introduction of oxygen vacancies to R-Co3O4 enhances its H2O2 adsorption ability and facilitates the decomposition of H2O2 to produce ·OH radicals, resulting in improved peroxidase-like activity. A simple and convenient colorimetric assay has been established based on the excellent peroxidase-like activity of R-Co3O4 for detecting H2O2 in concentrations of 1-30 μM with a detection limit of 4.3 × 10-7 mol/L (S/N = 3). Furthermore, the R-Co3O4-based colorimetric method was successfully applied to glucose detection in human serum samples, demonstrating its potential for application in complex biological systems.