N-butanol is a very competitive substitute fuel of gasoline due to its many advantages. In this study, a new reduced mechanism for n-butanol combustion simulation under engine conditions was developed based on the direct relationship graph with error propagation (DRGEP), sensitivity analysis and entropy production analysis methods. In the first stage, the DRGEP method and reaction pathway analysis (RPA) were used to remove unimportant species. In the second stage, the entropy production analysis method was utilized to eliminate insignificant reactions. In the last stage, for optimizing key reaction's rate constant and accurately reproducing the auto-ignition behavior, sensitivity analysis method was used to identify the important reactions. The proposed combined reduction approach successfully reduced the detailed n-butanol mechanism (326 species and 1884 reactions) to a reduced mechanism containing 75 species and 285 reactions. Extensive validations were performed by comparing experimental data of shock tube ignition delay times, jet-stirred reactor (JSR) species profiles, laminar flame speed and premixed laminar flame species profiles with modeled results. The reduced mechanism was also validated by comparing with the detailed mechanism in a single-zone engine model. High accuracy is demonstrated in wide ranges of temperature (T= 700-1500 K), pressure (p = 1-80 bar), and equivalence ratio (Phi = 0.5-2.0), which indicated that the present n-butanol reduced mechanism can well predict the experimental data. (C) 2016 Elsevier Ltd. All rights reserved.