One potential consequence of global climate change and rapid changes in land use is an increased risk of flooding. Proper understanding of floodwater infiltration thus becomes a crucial component of our preparedness to meet the environmental challenges of projected climate change. In this paper, we present the results of a long-term infiltration experiment performed on fractured ash flow tuff. Water was released from a 3 × 4 m² infiltration plot (divided into 12 square subplots) with a head of ∼0.04 m, over a period of ∼800 days. This experiment revealed peculiar infiltration patterns not amenable to current infiltration models, which were originally developed for infiltration into soils over a short duration. In particular, we observed that in part of the infiltration plot, the infiltration rate abruptly increased a few weeks into the infiltration tests. We suggest that these anomalies result from increases in fracture permeability during infiltration, which may be caused by swelling of clay fillings and/or erosion of infill debris. Interaction of the infiltration water with subsurface natural cavities (lithophysal cavities) could also contribute to such anomalies. This paper provides a conceptual model that partly describes the observed infiltration patterns in fractured rock and highlights some of the pitfalls associated with direct extension of soil infiltration models to fractured rock over a long period.