Experimental observations and theoretical studies over the last 10 years or so have demonstrated that flow channeling or preferred flow paths is a common phenomenon in fractured rocks. The reason it has come to the forefront of scientific investigation is the recent interest in predicting solute transport in geological media as part of safety assessment of geologic isolation of nuclear or toxic wastes. Solute transport is much more sensitive to medium heterogeneity than is temperature or pressure. In this paper, experimental observations of tracer transport over distances ranging from centimeters to hundreds of meters are reviewed, and theoretical efforts to explain or model these observations are summarized. Processes that may explain some of the experimental observations without the use of flow-channeling models are discussed. The paper concludes with a discussion of the implications of flow channeling on the practical problems related to contaminant transport in geologic systems. ¿ 1998 American Geophysical Union