The paper aims to fill the knowledge gaps in understanding hydrologic regime under climate change by using the newest IPCC assessment results .Generally the paper is well written with clear objectives. The authors use the routine methodology well developed in the literature, i.e., doing the hydrological modelling forced by ensembles of GCMs outputs. The results are clearly stated, with obvious difference from previous assessment with CMIP3. They have important implications for water resources management as well as disaster prevention.

Generally I agree with the Referees’ comments, and the authors should adequately address all the comments including the following ones:

(1) Similar with the Referee #1, one of my major concern is the uncertainty comparison between CMIP5 and CMIP3. The authors should highlight improvement made by the new study in a solid method.
(2) Hydrological model is important for the study. A review of relevant hydrologic researches in Mekong area is preferred, especially in terms of model skill. A related comment is the uncertainty of hydrologic modelling. As we can see from the FDC in Figure 2, the high flows and low flows are not sufficiently replicated. How can you say Q5 and Q95 indices are all within acceptable ranges. It requires more supports from literatures. Also, how this uncertainty influences the evaluation results needs more discussions.
(3) L21-23 Pg20: dam failure is an extreme case for dam operation. This ‘extreme’ is not the same ‘extreme’ in the paper title. The Q5 extreme is usually related to the flood regulation with the aim to protect the downstream area, while the extremes leading to dam failure often mean Q0.01 (1000-year return period) or even bigger events (e.g., 10000-year return period). The paper does not talk about these extreme extremes, so the discussion and conclusion about the dam failure is not in the proper context.
(4) Ln19-20 Pg6: South-West or South-East?

As suggested by the reviewers and reviewed by myself, I would like to say the revised manuscript improves significantly. However, still some concerns raised. I suggest the authors further revise the manuscript according to the new comments.