The goal of this exploratory research project is to control the pollutant emissions of diffusion flames by modifying the air infusion rate into the flame. The modification was achieved by installing a cascade of venturis around the burning gas jet. The basic idea behind this technique is controlling the stoichiometry of the flame through changing the flow dynamics and rates of mixing in the combustion zone with a set of venturis surrounding the flame. A natural gas jet diffusion flame at burner-exit Reynolds number of 5100 was examined with a set of venturis of specific sizes and spacing arrangement.The thermal and composition fields of the baseline and venturi-cascaded flames were numerically simulated using CFD-ACE+, an advanced computational environment software package. The instantaneous chemistry model was used as the reaction model.The concentration of NO was determined through CFD-POST, a post processing utility program for CFD-ACE+. The numerical results showed that, in the near-burner, midflame and far-burner regions, the venturi-cascaded flame had lower temperature by an average of 13%, 19% and 17%, respectively, and lower CO 2 concentration by 35%, 37%and 32%, respectively, than the baseline flame. An opposite trend was noticed for O 2 concentration; the cascaded flame has higher O 2 concentration by 7%, 26% and 44%, in average values, in the near-burner, mid-flame and far-burner regions, respectively, than in the baseline case. The results also showed that, in the near-burner, mid-flame, and farburner regions, the venturi-cascaded flame has lower NO concentrations by 89%, 70%and 70%, in average values, respectively, compared to the baseline case.The numerical results substantiate that venturi-cascading is a feasible method for controlling the pollutant emissions of a burning gas jet. In addition, the numerical results were useful to understand the thermo-chemical processes involved. The results showed that the prompt-NO mechanism plays an important role besides the conventional thermal-NO mechanism. The computational results of the present study need to be validated experimentally 4