This paper aims to measure the impact of HSR on urban land growth and to explore how the impact varies between different types of cites. Since HSR route planning decisions are influenced by the economic status of cities en route, whether a city gets an HSR station is likely to be endogenous to its economic growth and land development rate. To address this endogeneity, we adopt a Two-period Panel Data Instrumental Variable model using post road network in the Ming Dynasty and locations of military bases as instrumental variables. Results show that being connected to the HSR network on average leads to a 11.2% faster rate of growth of urban built-up areas. More interestingly, we document substantial heterogeneity in this impact based on the size and location of cities. Causes and mechanism for such heterogeneity are also elaborated, together with the policy and planning implications.