Rainfall-simulation experiments have been carried out on a series of plots ranging in size from 1 m² to c 500 m² in order to observe process and flux-rate changes resulting from the replacement of the dominant vegetation type from grassland to shrubland in the southwest. Results have demonstrated variations in infiltration rates, flow, hydraulics, splash and interrill erosion rates and nutrient transport rates. Furthermore, the shrubland areas develop rills, which are responsible for significant increases in overall erosion rates. The small-plot experiments allow the definition of controlling factors on the processes, and highlight the importance of vegetation controls. Although the small-plot approach has a number of significant advantages, it also has a number of disadvantages, which are discussed in detail. Some of these problems can be overcome with a careful consideration of experimental design. It is argued that plot-scale studies play an important part in improving our understanding of complex, open systems, but need to be integrated with other approaches such as the monitoring of natural events and computer modelling so that mutually consistent understandings of complex, ecohydrological systems can be achieved.