A numerical multicomponent reactive transport model described fully in Steefel and Lichtner (1998)[Steefel, C.I., Lichtner, P.C., 1998. Multicomponent reactive transport in discrete fractures, I. Controls on reaction front geometry. J. Hydrol. (in press)] is used to simulate the infiltration of hyperalkaline groundwater along discrete fractures at Maqarin, Jordan, a site considered as a natural analogue to cement-bearing nuclear waste repositories. In the Eastern Springs area at Maqarin, two prominent sets of sub-parallel fractures trending NW–SE are approximately perpendicular to the local water table contours, with the slope of the water table indicating north-westward flow. Extensive mineralogic investigations [Alexander W.R. (Ed.), 1992. A natural analogue study of cement-buffered, hyperalkaline groundwaters and their interaction with a sedimentary host rock. NAgrA Technical Report (NTB 91-10), Wettingen, Switzerland; Milodowski, A.E., Hyslop, E.K., Pearce, J.M., Wetton, P.D., Kemp, S.J., Longworth, G., Hodginson, E., and Hughes, C.R., 1998. Mineralogy and geochemistry of the western springs area. In: Smellie, J.A.T. (ed.), 1998: Maqarin Natural Analogue Study: Phase III. SKB Technical Report TR98-04, Stockholm, Sweden] indicate that the width of intense rock alteration zone bordering the fractures changes from about 4 mm at one locality (the M1 sampling site) to approximately 1 mm 100 m to the north-west in the flow direction (the M2 site), suggesting a lessening of alteration intensity in that direction. Using this information, the dimensionless parameter δ v/φ D′ (φ=porosity, D′=effective diffusion coefficient in rock matrix, δ=fracture aperture, and v=fluid velocity in the fracture) and measurements of the local hydraulic head gradient and effective diffusion coefficient in the rock matrix, a mean fracture aperture of 0.194 mm is calculated assuming the cubic law applies. This information, in combination with measured groundwater compositions at the Maqarin site, is used as input for numerical simulations of the hyperalkaline groundwater infiltration along fractures. The width of the alteration zones in the rock matrix bordering fractures is also used to constrain mineral dissolution rates in the field. The simulations predict that ettringite [Ca 6Al 2(SO 4) 3(OH) 12·26H 2O] with lesser amounts of hillebrandite and tobermorite (hydrated calcium silicates or CSH phases) will be the dominant alteration products forming at the expense of the primary silicates in the rock matrix and fracture, in agreement with observations at the Maqarin site. The simulations also come close to matching the pH of water samples collected along fractures at the M1 and M2 sites, with a fracture aperture of 0.22 mm giving the closest match with the pH data (within 13% of the value indicated by the rock matrix alteration widths). The simulations suggest two possible scenarios for the time evolution of the fracture–rock matrix system. Where rate constants for secondary mineral precipitation reactions are the same in both the rock matrix and fracture, the rock matrix tends to become completely cemented before the fracture. This results in a downstream migration of the hyperalkaline plume. In contrast, if rates are as little as one order of magnitude higher in the fracture than in the rock matrix, it is possible to seal the fracture first, thus causing the mineral zones to collapse upstream as a result of the reduction in fracture permeability. Sealing of fractures is observed at Maqarin and the simulations predict a mineral paragenesis in the fracture resulting from this scenario which is broadly compatible with field observations.