This paper reviews and analyses the major findings of research during the last 5 years in the Mechanical Engineering Department, Imperial College, on the chemistry of fuel-bound nitrogen in pulverized coal flames. The aim has been to lay the foundations for nitrogen-related laboratory research on p.f. flames of practical interest. The work has included both measurements and mathematical modelling. The experiments were conducted in both small-scale turbulent coal flames and large-scale laboratory flames (0.5 MW). The emphasis has been three-fold: 1. (1) Study of the initial stages of the devolatilization, nitrogen release and subsequent nitric oxide formation from pulverized coal particles injected into a bench-top turbulent flat flame; 2. (2) Identification of the combustion aerodynamic conditions in a large-scale pulverized-coal-fired laboratory furnace which favour reduced NO formation/emission, and determination of formation mechanisms of nitrogen oxides. To this end, measurements were made on two aerodynamically distinct industrial-type burners to assess the effect of furnace operating conditions and coal particle size distribution on NO formation and emission. Detailed in-flame measurements of nitrogen oxides, intermediate nitrogenous species, major gas species, gas temperature and particle composition were undertaken. 3. (3) Development and validation of an NO post-processor coupled to a 2D mathematical model (FAFNIR) of pulverized coal combustion. The NO post-processor is based on available empirically based, simplified kinetic schemes. The modelling encompasses the separate contributions to fuel-NO formation of the volatiles and the char as well as NO reduction by the char. The most significant conclusions to date include the following: 1. (1) A low-NO x burner has been developed, based on the knowledge gained on the effect of particle trajectories in the near-burner region on NO formation and reduction, which can successfully reduce NO emissions from 600 to 280 vpm without affecting flame stability and combustion efficiency; 2. (2) N 2O survives in the fuel-rich region where HCN, NH 3, and NO are present. There is no evidence of a post-flame N 2O formation ‘window’ between 877 and 1227 °C as suggested in some earlier studies; 3. (3) The NO predictions are generally in good agreement with a wide range of experimental data, except for the near-burner region of the low-NO x burner, where the model somewhat over-predicts the combustion reactions. This deficiency in the parent code, which appears to be shared by all other codes, is due to the poor predictions of the gas temperature and oxygen concentrations in the near-burner region.