Previous studies have found an inverse (or negative) correlation between urban population density and per capita emissions from land transport. In contrast, this paper finds a positive relationship between per capita CO2 emissions from transport and population density using a dataset of over 200 cities from 28 countries. This positive relationship holds when a range of variables are accounted for and the specification of the regression analysis captures the distinction between country level differences, high/low emission intensity or city specific fixed effects. Separating the cities into two groups based on the clustering that occurs on either side of a crucial point of three tonnes of CO2 emissions per capita highlights the peculiarity of the higher emission intensity of North American cities. Rather than finding a consistent relationship across all cities, this paper finds that cities in North America are distinct from those located in other countries and that the estimated relationship between urban population density and emissions from transport is different across the two groups of countries. The results of this paper have consequences for policy prescriptions that are related to previous results that find that a reduction in per capita emissions tends to occur with an increase population density.