AbstractPrevious studies have shown that road networks have a wide range of effects on land use and ecological processes, which result in both direct and indirect impacts on ecosystems. However, some of these studies employed predictive inference approaches to identify the causal relationships; we argue that these methods may generate biased conclusions. In this study, taking as the study area the biggest central Chinese megacities where land-use change was massive over the past decades, we investigate the causal relationships between the city’s road network and its landscape ecological risk using different quasi-experimental designs. Two different geographic scopes of road effects as well as the intensity of the effects are examined. For the years 2005, 2012, and 2019, we measure the distance to the nearest road, road density, and kernel density as well as landscape ecological risk at a 5-km2 block scale. When applying correlation analyses and ordinary least squares regressions without experimental settings, these data indicate that blocks closer to the road and having higher road density, or kernel density, are associated with higher landscape ecological risk values. Yet, difference-in-difference estimated results suggest new roads built during the 2005–2012 period did not have an economically significant effect on subsequent landscape ecological risk.