Urban sinkholes can cause subsidence damage to transportation infrastructures, building demolitions and even casualties when they occur suddenly. It is hence essential to identify their forming factors and analyze their spatial distribution features, so as to pave the way for sinkhole risk assessment and mitigation for urban sustainable development. Investigation of urban sinkholes poses a significant challenge because of the multi-factor’s confusion, as well as the spatial uncertainties in susceptibility mapping due to urban renewal frequently. To address these issues, a comprehensive urban sinkhole risk assessment framework was proposed, and Shenzhen city in South China was selected as a case study area, where a lot of sinkholes occurred along with intensive human construction activities in recent years. Initially, the sinkhole susceptibility map was drawn using the grey relation analysis (GRA) method, incorporating the spatial information database of meteorology, geology, metro and pipeline systems, and then both the sinkhole density map and the time-varying land subsidence monitoring database were used to verify the accuracy of the sinkhole susceptibility map. Furthermore, the sinkhole risk assessment map in transportation system was subsequently drawn based on the susceptibility map, where the Interferometric Synthetic Aperture Radar (InSAR) and the advanced geophysical exploration technologies were employed to probe the potential sinkholes in critical risk areas. The conclusions reveal that the sinkhole susceptibility map based on GRA method demonstrates a promise in identifying the likelihood of urban sinkholes, and the risk assessment framework can be employed for the prevention and remediation of urban sinkholes, which contributes to an economical and efficient scheme for sinkhole detection, as well as transportation system resilience assessment at a city-wide scale.