While diesel engines are arguably superior to any other power-production device for the transportation sector in terms of efficiency, torque, and overall driveability, they suffer from inferior performance in terms of noise, NO, and particulate emissions. The majority of particulate originates with soot particles which are formed in fuel-rich regions of burning diesel jets. Over the past two decades, our understanding of the formation process of soot in diesel combustion has transformed from inferences based on exhaust measurements and laboratory flames to direct in-cylinder observations that have led to a transformation in diesel engine combustion. In-cylinder measurements show the diesel spray to produce a jet which forms a lifted, partially premixed, turbulent diffusion flame. Soot formation has been found to be strongly dependent on air entrainment in the lifted portion of the jet as well as by oxygen in the fuel and to a lesser extent the composition and structure of hydrocarbons in the fuel. Soot surviving the combustion process and exiting in the exhaust is dominated by soot from fuel-rich pockets which do not have time to mix and burn prior to exhaust valve opening. Higher temperatures at the end of combustion enhance the burnout of soot, while high temperatures at the time of injection reduce air entrainment and increase soot formation. Using a conceptual model based on in-cylinder soot and combustion measurements, trends seen in exhaust particulate can be explained. The current trend in diesel engine emissions control involves multi-injection combustion strategies which are transforming the picture of diesel combustion rapidly into a series of low temperature, stratified charge, premixed combustion events where NO, formation is avoided because of low temperature and soot formation is avoided by leaning the mixture or increasing air entrainment prior to ignition. (c) 2006 Elsevier Ltd. All rights reserved.