Abstract Cities are highly vulnerable to climate change, yet the interactions between urban and regional climate remain insufficiently understood, especially over climate changes timescales and when comparing cities globally. Therefore, this study assesses the capabilities of two CORDEX-CORE regional climate models (REMO and RegCM) in representing urban areas globally, focusing on land-surface characteristics and urban heat island (UHI) evaluation. Despite their relatively coarse resolution (~25 km), the two models can capture urban imprints of large cities. RegCM, with a single-layer urban canopy parameterization, represents the UHI, especially at night. REMO tends to underestimate nighttime UHI due to its simple bulk urban scheme. Across models, impervious surface areas are consistently underestimated, with notable geographic imbalances across the world. Going forward, regional climate model simulations for cities require both enhanced urban parameterizations and the integration of refined urban land-use data.