The lattice Boltzmann method (LBM), a mesoscopic method between the molecular dynamics method and the conventional numerical methods, has been developed into a very efficient numerical alternative in the past two decades. Unlike conventional numerical methods, the kinetic theory based LBM simulates fluid flows by tracking the evolution of the particle distribution function, and then accumulates the distribution to obtain macroscopic averaged properties. In this article we review some work on LBM applications in engineering thermophysics: (1) brief introduction to the development of the LBM; (2) fundamental theory of LBM including the Boltzmann equation, Maxwell distribution function, Boltzmann-BGK equation, and the lattice Boltzmann-BGK equation; (3) lattice Boltzmann models for compressible flows and non-equilibrium gas flows, bounce back-specular-reflection boundary scheme for microscale gaseous flows, the mass modified outlet boundary scheme for fully developed flows, and an implicit-explicit finite-difference-based LBM; and (4) applications of the LBM to oscillating flow, compressible flow, porous media flow, non-equilibrium flow, and gas resonant oscillating flow. lattice Boltzmann method, numerical simulation, engineering thermophysicsThe lattice Boltzmann method (LBM), a mesoscopic numerical method based on the kinetic theory, has attracted the attention of many scholars at home and abroad and has been developed rapidly in both theory and applications in the past years. At the macroscopic level, LBM can be considered as a discrete method, while at the microscopic level it can be treated as a continuum method. In comparison with conventional numerical methods, kinetic features of LBM enable it to be more effective for simulating complex flows, such as flow in porous media, suspension flow, multiphase flow, and multi-component flow. Its main advantages can be briefly summarized as follows: nearly ideal amenability to parallel computing, complex boundary conditions can be easily formulated in terms of elementary mechanics rules, and simple programming.LBM historically originated from the lattice gas automata method which has been plagued by several diseases: (1) violation of the Galilei invariance; (2) explicit dependence of equation of state on velocity; (3) noise due to Boolen variables; and (4) complicated collision operator.In order to get rid of the noise in the lattice gas automata method, McNamara and Zanetti [1] introduced the first lattice Boltzmann model in 1988, in which Boolen fields were replaced by continuous distributions over the lattices, and the Fermi-Dirac distribution was used as the equilibrium distribution function. In 1989, Higuera and Jimenez made an important simplification of the lattice Boltzmann method, in which they proposed to use a linearized collision operator by assuming that the distribution was close to its local equilibrium state [2,3] . Subsequently, the Bhatnagar-Gross-Krook (BGK) collision operator was independently introduced into LBM by