The paper describes a set of drought projections for Australia developed by dynamically downscaling CMIP6 global climate models. A possible range of future drought conditions is considered that span multiple emissions pathways and model configurations. Future drought conditions are described through event frequency, duration, spatial extent and time spent in drought, in terms of changes in two commonly used drought metrics. The paper is well-structured, the information is clearly presented and the key results appropriately discussed. I recommend publication after minor revisions, which I describe below.

One key issue is the use of SPEI. This metric, being the difference between precipitation and potential evapotranspiration, is intended to reflect the atmospheric water balance and thereby give a complementary view to SPI-based drought. However, the use of potential evapotranspiration in the calculation of SPEI makes SPEI unrealistic in many water-limited parts of Australia, where actual evapotranspiration does not approach the potential upper limit. So, any projected worsening of PET-related conditions is merely an indication of an increase in atmospheric demand for moisture, rather than a conclusive reduction in water stores. I suggest this issue is more adequately discussed in the paper, including the implications in the interpretation of SPEI-based projections of drought.

The second key issue is the lack of attention given to the uncertainty of the projections. While using a multi-model ensemble and multiple emissions pathways goes some way to addressing uncertainty, the drought projections should be presented along with quantified uncertainty estimates. Moreover, the issue of uncertainty propagation from GCM through to downscaling technique to RCM was not addressed.

The final key issue is that one of the most crucial findings of the study needs to be made more prominent. The results show that more time is projected to be spent under extreme conditions, both wet and dry, and less time under ‘normal’ conditions, for some parts of Australia (Table 3). This result should be made more prominent, for example by featuring in the abstract. This result is important because it suggests that the combination of projected changes in the climate system is shifting the dial towards more extreme climatic conditions and motivates future research in understanding the physical processes responsible for the shift.

Minor comments:

• L84: this is a bit of a throw away line. I suggest turning this around by stating that since RCMs have been shown to estimate regional rainfall features with higher precision than GCMs, RCMs are more appropriate to study drought on the regional scale.
• Inter-model variability (Figure 11) is shown to be higher for SPEI and some drought characteristics. Can an explanation be offered for why this is? What are the implications of this variability on the interpretation of future drought changes?