[Purposes] To address the issues of insufficient rationality in the allocation of subjective and objective weights, risk uncertainty, and difficulties in risk characterization in road collapse risk assessment, this study aims to establish a scientific and reliable risk assessment model to provide technical support for road collapse prevention and control. [Methods] An evaluation system was constructed by selecting 11 indicators from four dimensions: water-related pipelines, construction activities, natural environment, and road conditions. Subjective weighting method (G1 method) and objective weighting method (EW method) were employed to calculate subjective and objective weights respectively, which were then combined and optimized based on game theory principles. The cloud model was utilized to quantify the randomness and fuzziness of risks, and risk levels were determined through expectation value comparison. [Findings] Against the backdrop of an engineering case, comparative analysis revealed that the game theory-based combined weighting significantly enhanced the differences in importance among indicators compared to the traditional additive synthesis method, and markedly improved the deviation issues of minor indicators compared to the multiplicative normalization method. The risk levels of the four evaluation units were classified as Level I (low risk), Level II (medium risk), Level III (high risk), and Level IV (extremely high risk), respectively. The risk levels of the evaluation units exhibited a high degree of agreement with the distribution and risk levels of underground defects detected by ground-penetrating radar, with an evaluation accuracy rate of 77.8%. [Conclusions] The game theory-based combined weighting effectively balances the subjective experience limitations of the G1 method and the data dependency flaws of the EW method, avoiding the drawbacks of traditional combined weighting methods. The cloud model objectively characterizes risk boundaries through three parameters: expectation value, entropy, and hyper-entropy. This method can serve as a reliable tool for dynamic assessment of urban road collapse risks.