This study presents an important national-scale assessment of flood risk in India. The methodology is adequate and the manuscript is well written. I thus recommend its acceptance in HESS after minor review.

My only concern is that it is not clear throughout the text how reservoirs were considered in the study. In some parts of the methodology it is argued that the model is calibrated with dam storage data, and that the model considers the role of dams. Yet later it becomes more clear that actually the C-ratio was used, which is a proxy of storage effects; i.e. dams were actually not simulated by the model. Please clarify it across the text. Also, C-ratio is a proxy of storage effects and has some limitations. RoR dams can sometimes have high C-ratio but low impacts on downstream hydrology. Please add something about it to the discussion.

Minor comments:

Line 15-16 This sentence is too vague, please rephrase it: 'Sub-basins in the Ganga and Brahmaputra River basins witnessed the greatest flood extent during the worst flood in the observational record.'

L. 66 This reference on the number of dams is from 2000. Is this number updated?

L. 100 GSW cannot estimate flooded vegetation and is usually limited for flood events because of cloud influence. Furthermore, your analysis is for 1901-2020, yet GSW covers only 1985-present. Please clarify the use of GSW in the text.

L. 121 At this stage, it seems that you've simulated the dam impact on downstream hydrology, but later in paragraph 138-148 you mention that the dam potential impact on hydrology was calculated with the C-ratio. Please clarify it throughout the text.

L. 212-213 To illustrate the fact that ' the worst flood in the same year did not affect all the sub-basins within a river basin', it would be very interesting to see a heat map of the drainage network colored according to the year of the worst flood in the century. It would be easy to make it based on the model outputs.

L. 221 It would be important to present in the larger map the locations of the sub-basins (a to j)

L. 234-234 Are there any references (especially in grey literature) that could be used to support your findings of when the worst flood occurred in the different sub-basins?

L. 279 Change to uppercase: 'C-ratio'

L. 338 Typo: 'Hirabayashi et al. (2013)'

L. 336 I missed in the discussion some comments about which mitigation measures are being undertaken by India at multiple government levels.

The study uses a modeling approach aiming to assess the flood risk at the sub-basin scale in India and the impact of reservoirs on the flood risk. My recommendation is rejection due to several significant issues in the methodology that seem inadequate for addressing the posed questions. Further justification is also necessary for the conclusions made. Here are my main concerns:

• The understanding of flood risk seems to focus too heavily on the worst flood event in history. To understand flood risk, it requires examination of a large number of flood events over a range of conditions and incorporating uncertainties.
• I question the suitability of a large-scale model like H08-CaMaFlood for flood risk assessment, which typically requires higher-resolution models that can accurately capture local topography and features. Given the shown substantial bias in simulated flood occurrences, I am unconvinced of the model's efficacy in predicting flood water depth at the event scale. The downscaling approach, which scales simulated flood depth from a 0.1 degree to a 200 m resolution within CaMa-Flood, compounds this uncertainty. Ultimately, the rationale behind the choice of CaMa-Flood for localized flood risk assessment is unclear to me, as its resolution seems too coarse for the purpose.
• The authors' claim of an acceptable model skill is unconvincing to me. For river flooding, they set a NSE threshold of 0.5, which is questionable since a score of 0.6 is generally considered the minimum for model adequacy. Even then, some stations fail to meet this lowered threshold. There is also a lack of flood-relevant metrics, such as bias in peak discharge of flood events. I would suggest evaluating the worst flood event selected as well. Concerning flood inundation modeling, it would be beneficial if flood extent data were used to evaluate the model's skill. With respect to flood occurrences, I noted previously that the bias seems significant even before the application of downscaling. Despite these evident issues, no discussions on the uncertainties present in this study are included. This omission casts further doubt on the reliability of the results.
• Concerning flood inundation modeling, it would be beneficial if flood extent data were used to evaluate the model's skill. With respect to flood occurrences, I noted previously that the bias seems significant even before the application of downscaling. Despite these evident issues, no discussions on the uncertainties present in this study are included. This omission casts further doubt on the reliability of the results and necessitates a comprehensive review of the methodology. The use of the C-ratio to assess the role of reservoir operations in flood risk is confusing. The C-ratio, defined as the ratio of a reservoir's total maximum storage capacity to the mean annual discharge at the sub-basin outlet, is essentially a constant that doesn’t account for variability in reservoir outflow resulting from operations serving different objectives. The mean annual discharge also seems irrelevant when examining a record flood event at a much shorter timescale. Consequently, I find the results based on C-ratio to be lacking in significance.
• Certain fundamental details that could aid in interpreting the results are missing, such as a clear definition of how a flood event is defined.