Carroll et al. (2009) state that the United States Geological Survey (USGS) Death Valley Regional Flow System (DVRFS) model, which is based on MODFLOW, is “conceptually inaccurate in that it models an unconfined aquifer as a confined system and does not simulate unconfined drawdown in transient pumping simulations.” Carroll et al. (2009) claim that “more realistic estimates of water availability” can be produced by a SURFACT-based model of the DVRFS that simulates unconfined groundwater flow and limits withdrawals from wells to avoid excessive drawdown. Differences in results from the original MODFLOW-based model and the SURFACT-based model stem primarily from application by Carroll et al. (2009) of head limits that can also be applied using the existing MODLOW model and not from any substantial difference in the accuracy with which the unconfined aquifer is represented in the two models. In a hypothetical 50-year predictive simulation presented by Carroll et al. (2009), large differences between the models are shown when simulating pumping from the lower clastic confining unit, where the transmissivity is nearly two orders of magnitude less than in an alluvial aquifer. Yet even for this extreme example, drawdowns and pumping rates from the MODFLOW and SURFACT models are similar when the head-limit capabilities of the MODFLOW MNW Package are applied. These similarities persist despite possible discrepancies between assigned hydraulic properties. The resulting comparison between the MODFLOW and SURFACT models of the DVRFS suggests that approximating the unconfined system in the DVRFS as a constant-saturated-thickness system (called a “confined system” by Carroll et al., 2009) performs very well.