This work reviews the runoff generation process in forests. A survey of the delivery mechanisms of hillslope runoff and the difficulties of incorporating some of them in recent physically based models is considered. The research challenge in reconciling the results from recent stream hydrogeochemistry studies with the results from previous hillslope hydrometric experiments is also highlighted. 'Physically based' runoff process models for application to forest land management problems are then summarised. The work concludes by proposing a new, intensive phase in experimental hydrology which should support the continued development of new algorithms in runoff process modelling. Particular attention should be given to tropical forests and the need for additional hillslope hydrology research to address the important issues associated with hillslope management and conversion.