Various thermal-hydrologic models have been developed to simulate thermal-hydrologic conditions in emplacement drifts and surrounding host rock for the proposed high-level nuclear waste repository at Yucca Mountain, Nevada. The modeling involves two-phase (liquid and gas) and two-component (water and air) transport in a fractured-rock system, which is conceptualized as a dual-permeability medium. Simulated hydrologic processes depend upon calibrated system parameters, such as the van Genuchten alpha and m, which quantify the capillary properties of the fractures and rock matrix. Typically, these parameters are not calibrated for strongly heat-driven conditions, i.e., conditions for which boiling and rock dryout occur. The objective of this study is to modify the relationship between capillary pressure and saturation, P (c)(S), for strongly heated conditions that drive saturation below the residual saturation (S -> 0). We offer various extensions to the van Genuchten capillary-pressure function and compare results from a thermal-hydrologic model with data collected during the Drift-Scale Test, an in situ thermal test at Yucca Mountain, to investigate the suitability of these various P (c) extension methods. The study suggests that the use of extension methods and the imposition of a capillary-pressure cap (or maximum) improve the agreement between Drift-Scale Test data and model results for strongly heat-driven conditions. However, for thermal-hydrologic models of the Yucca Mountain nuclear waste repository, temperature and relative humidity are insensitive to the choice of extension method for the capillary-pressure function. Therefore, the choice of extension method applied to models of drift-scale thermal-hydrologic behavior at Yucca Mountain can be made on the basis of numerical performance.