Urban road infrastructure is crucial in affecting air pollution. Yet, little is known about the roles played by road width vs road length. This paper attempts to fill this gap by estimating the effects of road infrastructure on PM10 using city-level data from China. Our robust modeling results show that the road density index, defined as the ratio of road surface area to city territory size, is negatively correlated with PM10. More importantly, when the road width and length components of the road density variable are separately included in the regression models, the width is found to be significantly and negatively correlated with PM10, whereas the correlation with the length variable is insignificant. This is expected as increases in road width can help mitigate congestion, improve fuel efficiency, and thus reduce emissions. On the contrary, extending roads to new or unconnected areas is likely to bring more vehicles onto the roads and extend average driving time. Our findings appeal for careful consideration of the trade-off between road width (lane numbers) and road length when planning and constructing urban road infrastructure.