We used the mobility-emission methods (https://data.d4science.org/ctlg/ResourceCatalogue/mobility-emissions) to perform experiments on real GPS trajectories describing 433,272 trips from 14,907 private vehicles moving in Greater London, Rome, and Florence throughout January 2017. We find that, for all three cities, emissions distribute across vehicles in a heterogeneous way: a few vehicles, that we call gross polluters, are responsible for a tremendous amount of emissions. At the same time, most of them emit significantly less. Also, a few grossly polluted roads suffer from a significant quantity of emissions, while most of them suffer significantly fewer emissions. Both the distributions of emissions across the vehicles and across the roads are well approximated by heavy-tailed distributions (mainly a truncated power law or a stretched exponential, with parameters changing with the city). We also investigated the relationship between a vehicle's emissions and its mobility behaviour, from one side, and between the emissions suffered by a road and its network features, discovering that gross polluters tend to be more regular and predictable in their mobility than low-emitting vehicles, and confirming that the most polluted roads are the ones that fall more frequently on the shortest paths connecting two nodes of the network. Finally, as reducing emissions is a growing concern for cities, and estimating the impact of policies targeting vehicles to reduce their footprint on the city’s environment is crucial, we investigate the impact that (i) the vehicles’ electrification and (ii) the home working have on the total amount of emissions and the distribution of emissions across the roads. We find that, for example, the electrification of just the top 1% gross polluters moving in Rome would lead to the same reduction of the CO2 emitted overall as electrifying 10% random vehicles, and that the remote working of the top 1% gross polluters would lead to the same reduction reached if they were ∼4% random vehicles. Similar results hold for the other cities, even if with slightly lower numbers for London.