The application of rolled erosion control systems (RECSs) to disturbed hillslopes has signi®cantly increased during the 1990s. Erosion-related research conducted on RECSs during the post-1990 period was dominated by black-box output studies with minimal attention to erosion process relationships. Additionally, many studies failed to incorporate suitable treatment replication, and in others in¯uential variables were not controlled between experiments making results dicult to interpret. However, there were some important advances made in laboratory rainfall simulation studies, controlled greenhouse studies, and rainfall simulation studies. The ®rst detailed studies attempting to link material properties of RECSs and their erosion control eectiveness were initiated; links between salient material properties of RECSs and erosion process dynamics (interrill and rill) were formally explored; detailed studies of the in¯uence of RECSs on hydrological and microclimatological variables were examined in both greenhouse and ®eld experiments; and an examination of temporal variations in runo and erosion were introduced into the RECS literature. Despite recent advances there are still a number of signi®cant gaps in our understanding. These gaps are exacerbated by lack of standard ®eld or laboratory testing protocols. Two of the outstanding questions which need further study include: What are the most critical properties of RECSs that decrease soil erosion and runo, but enhance growth of vegetation? What are the most eective organic or synthetic RECSs in controlling erosion on a variety of slopes and soil textures? It is argued that standardized protocols are needed, but these must include adequate replication and randomization. Additionally, there must be an emphasis on understanding the basic surface or near-surface processes in¯uenced by RECSs (i.e. erosional, hydrological, biological, and climatological).