Some conclusions from a recent numerical modeling study regarding the potential intensity of tropical cyclones are shown to be attributable to a relatively short model integration time. Three-dimensional simulations that are long enough to allow for a quasi-steady state of maximum intensity show a result consistent with theory, i.e., that maximum intensity is inversely proportional to surface drag coefficient. An alternative hypothesis for how tropical cyclones can intensify in numerical models in the absence of surface drag is also offered.Key Words: hurricane; surface exchange coefficient; numerical simulation In a recent numerical modeling study, Montgomery et al.(2010, hereafter MSN10) examined the sensitivity of tropical cyclone (TC) intensification to the surface drag coefficient. As part of their study, MSN10 compared their results to previous studies on maximum possible intensity (also known as potential intensity, PI). This comparison to studies on PI should have been done more carefully because there is a significant difference in methodology between MSN10's simulations and the methodology used in previous studies of PI; specifically, the length of the model integration is approximately a factor of 3 smaller in MSN10. The primary purpose of this Comment is to document this key difference, and to demonstrate how this different methodology can lead to different conclusions. I also offer an alternative hypothesis for how TCs can intensify in numerical models in the absence of surface drag.Tropical-cyclone PI is usually considered to be proportional to the surface exchange coefficient for enthalpy C k because, all else being equal, larger C k results in larger surface heat flux (which is the ultimate source of energy for TCs). Furthermore, PI is usually considered to be inversely proportional to the surface exchange coefficient for momentum C d because, all else being equal, larger C d (i.e. surface drag) results in larger kinetic energy dissipation. More specifically, some theoretical studies have determined that maximum tangential velocity V max should vary as follows:(1) (e.g. Emanuel, 2004, and references therein). It is important to note that Eq. (1) applies to the maximum possible tangential velocity for a specified environment (e.g. seasurface temperature, atmospheric moisture/temperature profile), assuming all else held fixed. Furthermore, Eq. (1) is derived from equations assuming steady flow (i.e. Eulerian time-tendency terms are negligible). Some support for Eq. (1) was found in numerical modeling studies by Emanuel (1995) and Bryan and Rotunno (2009). Using two different axisymmetric models (one based on gradient-wind balance and the other a primitive-equation model), Emanuel (1995) evaluated Eq.(1) by integrating these models 'for long enough that quasisteady mature model storms were achieved' (p. 3972). Using a different primitive-equation axisymmetric model, Bryan and Rotunno (2009) integrated numerical simulations for 12 days in order to achieve 'an approximately steady state ' (p. 1775)...