Many bays in the world are threatened by coastal hazards such as storm surge, river flood and tsunami.Since most of the existing studies have been focused on one or two of them, in this study, the assessment of coastal vulnerability caused by the three hazards was the research target. Inundation simulation is a widely used and straightforward way in coastal vulnerability assessments; however, it is computationally expensive, and considering an increase in the number of cases in multi-hazard analysis, an efficient method was proposed using an estimated overflow volume without computing inundation, which was validated by comparing with inundation simulation. It shows that when free overflow is dominant, this method is consistent with inundation simulation approach. Using Tokyo Bay as a study area, the efficient method was then applied to multi-hazard vulnerability assessment. By comparing the overflow volume maps and maximum anomaly distribution along the coasts for four types of hazards (worst storm surge; worst concurrent storm surge and river flood; worst concurrent storm surge, river flood and Tokai-Tonankai earthquake tsunami; worst concurrent storm surge, river flood and Tokyo inland earthquake tsunami), we investigated the characteristics of different types of hazards and identified the difference between single hazard and multi-hazards. The characteristic of overflow volume along the coasts is similar to that of maximum anomaly distribution, especially for only storm surge case, the multi-hazard case combining storm surge and river flood, and the multi-hazard case combining storm surge, Tokyo inland earthquake tsunami and river flood. However, for multihazard case combining storm surge, Tokai-Tonankai earthquake tsunami and river flood, only by the maximum anomaly distribution, it cannot reflect the real overflow volume condition. For only storm surge case and multi-hazard case combining storm surge and river flood, the head of the bay suffers the highest vulnerability while for multihazard cases combining storms surge, tsunami and river flood, the difference of vulnerability in the north and south of the bay is not significant. The difference of superposing method and concurrent method for computing multihazards was also compared. It was found that the linear superposing method tends to overestimate the total water elevation in coastal region; however, in the coasts where superposing method underestimates the multi-hazard anomalies, upgrading dikes needs to be considered by policymakers.