The purpose of this work is to study the transport of water (both the liquid and vapor phases) in tuff from the Topopah Spring welded unit under conditions expected in the near-field environment of a high-level nuclear waste container. A naturally fractured sample of Topopah Spring tuff (approximately 8 cm in diameter and 10 cm long) from Yucca Mountain at the Nevada Test Site, Nevada, was studied using, as a pore fluid, natural groundwater recovered from a well in which the principal producing horizon is the Topopah Spring Member. Confining pressure, sample temperature, and pore pressure were held at values that simulate expected in situ near-field conditions shortly after emplacement of the container. Results of this work are comparable with results from previous similar experiments on 2.54-cm-diameter samples of Topopah Spring tuff. During the more than 6-month experiment duration, water permeability decreased about 3 orders of magnitude. The most rapid measured permeability change (between about 600 and 100 μdarcy) occurred when the sample temperature was increased from room temperature to 89°C. Subsequently, water permeability decreased in a fairly monotonie manner to a value of about 1 μdarcy. This behavior probably resulted from a decrease in the aperture of the natural fracture, possibly reflecting transport and redeposition of silica (which was already present in the fracture). This process also resulted in a weak bonding of the sample halves. Previous experiments on two smaller naturally fractured Topopah tuff samples resulted in fracture healing presumably by the same mechanism. The degree of healing, indicated by bonded tensile strength and change in fluid permeability, seems to be sample dependent. The distribution of electrical resistivity indicates that the sample dehydrates and rehydrates nonuniformly. This is consistent with our findings on smaller samples. Computed impedance tomographs indicate that during each of the five subsequent dehydrations of the sample, water first left the matrix adjacent to the fracture surface, then escaped the sample through the fracture aperture. Drying then progressed into the matrix around the fracture. Rehydration of the dry sample was not quite the reverse of the dehydration process. Water initially entered the fracture along the fracture edge, and the matrix was then wetted progressively from these points. However, the matrix was also wetted along paths apparently unrelated to fracture flow.