An index which links the fuel composition to particulate matter emissions (the PN index) was developed and is here evaluated with model fuels in a single-cylinder optical-access spray-guided direct-injection engine; the model fuels have independent control of the double-bond content and volatility, as used in the index. This index is investigated in three different engines: a single-cylinder research engine, a V8 engine recently available in the market and a current-production supercharged V6 engine. A number of market gasolines were tested in all three engines, and the results follow the trends predicted by the particle number index. Imaging of the in-cylinder sprays shows that the composition of the model fuels affects the mixture homogeneity and their particulate matter emissions; in particular, the lack of a light end in the model fuel composition can lead to misleadingly low particle number emissions owing to improved mixture preparation which is unrepresentative of market fuels. The PN index was investigated in a Jaguar Land Rover V6 engine with five different fuels over a simulated New European Driving Cycle, and the results show that the index trends are followed. The emissions were evaluated from two fuels representing the EU5 reference-fuel specifications that has been developed using the particle number index to give a difference in particulate matter emissions. The results from these fuels show that a difference in the particle number emissions of a factor of about 2 can be seen at both stoichiometric conditions and rich conditions, for two fuels representative of the EU5 reference-fuel specifications. This follows trends predicted by the particle number index. This has important implications for policy makers and European Union legislation, where particle number emissions from gasoline vehicles are now regulated for the first time, as batch-to-batch variations in the fuel composition would result in different test results under the current legislation.