Laminar flame speeds of methanol/air mixtures at 338–398 K are measured by the heat flux method, extending the range of equivalence ratio up to 2.1. And a new optimized methanol mechanism with 94 reactions is proposed by using the particle swarm algorithm, adjusting 20 Arrhenius pre-exponential factors in their uncertainty domains. The optimized model is compared with eight methanol combustion mechanisms and experimental data published in recent years, covering a wide range of initial temperatures (298–1537 K), pressures (0.04–50 atm) and equivalence ratios (0.5–2.1). The results show that the optimized mechanism not only improves the accuracy of ignition delay time with rapid compression machine at low temperature but also moderately improve the description of laminar flame speed in lean and stoichiometric conditions. Meanwhile, the optimized model significantly enhances the prediction accuracy of CH₃ and CH₂O radical, and perfectly captures the evolution trend of HCO radical in laminar flat flame. Overall, the optimized mechanism provides the best overall description of the currently available measurements, leading to more accurate and comprehensive prediction of ignition delay time, laminar flame speed and species concentration.