This introduction to the second annual review issue of Hydrological Processes tries to put the collection of papers on preferential flows and travel time distributions into a more general context of testing models as hypotheses about how catchment systems function. It is suggested that, because of the possibilities of non-stationary and epistemic errors in both data and models, such tests could be carried out within a rejectionist limits-of-acceptability framework. The principles and difficulties of hypothesis testing within these particular research areas are discussed. An important point to take from this discussion is that the use of a formal testing framework, and the consequent rejection of models as hypotheses after allowing for uncertainties in the data, is the starting point for developing better theories and data sets.