Abstract We revisit the radiative neutron capture reaction 16O(n,γ)17O of astrophysical interest, based on the new reevaluated cross-section data. Several potentials are proposed to predict direct capture cross sections. The contributions from single-particle resonances to total capture cross section are quantitatively considered in Breit–Wigner formalism, taking into account the interference term between direct capture and resonant cross sections, which is crucial for the description of the behavior around the resonance energies. A new cross section is achieved based upon χ 2-fittings for optimized resonance parameters using Minuit code, and it has a largely improved agreement with updated experimental data. Statistical errors are also evaluated for the total and Maxwellian-averaged cross sections. It is confirmed that the direct captures dominate the total cross sections; however, resonant contribution also becomes progressively more important as the energy increases to 100 keV. Resonance contribution can increase the reaction rates for energy region 50 keV < E < 100 keV by 5% ± 5%–25% ± 5%, and around 8% ± 5% in comparison with KADoNiS v0.3 rate and the latest data evaluations, respectively. We show a detailed propagation of the uncertainty in the 16O(n,γ) reaction rate to abundances of nuclei, including s-nuclei during the weak s-process with a multi-zone nuclear network calculation. Although an enhanced rate of 16O(n,γ) diminishes the s-process efficiency in the 25 M ☉ stellar model adopted from the Modules for Experiments in Stellar Astrophysics, it can lead to larger abundances of neutrons as well as 22Ne in the late epoch of C burning.