In this paper, detailed computational study is presented which helps to understand and improve the fuel-air mixing in a new direct-mixture-injection two-stroke engine. This new air-assisted injection system-based two-stroke engine is being developed at the Indian Institute of Science, Bangalore over the past few years. It shows the potential to meet emission norms such as EURO-II and EURO-III and also deliver satisfactory performance. This work proposes a comprehensive strategy to study the air-fuel mixing process in this engine and shows that this strategy can be potentially used to improve the engine performance. A three-dimensional compressible flow code with standard k−2 turbulence model with wall functions is developed and used for this modeling. To account for the moving boundary or piston in the engine cylinder domain, a non-stationary and deforming grid is used in this region with stationary cells in the ports and connecting ducts. A flux conservation scheme is used in the domain interface to allow the in-cylinder moving mesh to slide past the fixed port meshes. The initial conditions for flow parameters are taken from the output of a three-dimensional scavenging simulation. The state of the inlet charge is obtained from a separate modeling of the air-assisted injection system of this engine. The simulation results show that a large, near-stoichiometric region is present at most operating conditions in the cylinder head plane. The state of the in-cylinder charge at the onset of ignition is studied leading to a good understanding of the mixing process. In addition, sensitivity of two critical parameters on the mixing and stratification is investigated. The suggested parameters substantially enhance the flammable proportion at the onset of combustion. The predicted P - θ from a combustion simulation supports this recommendation. EF/proceedings. Copyright © 2005 by ASME.