The runoff threshold (P0) is one of the key parameters in the study of catchment hydrological response. It is fundamental in the analysis of rainfall-runoff conversion processes, for estimating water budgets and for describing the genesis of hortonian overland flow, as it provides information on the precipitation losses within the fluvial systems, from the start of the episode until runoff is produced. There are several methods for estimating this parameter. In the case of small semiarid and Mediterranean basins, complex models are not suitable due to the specific conditions of these environments as well as the quantity of data required for the models. In these environments, empirical methods have demonstrated to be more effective. One of the most used models in Spain is that developed by the US Soil Conservation Service (SCS) in 1972, and modified by Témez (1978, 1991). This is a simple, easy to apply model, which uses slope values, vegetation, land use and soil hydrological group data. In spite of the generalized application of the model and its multiple adaptations, there are still many uncertainties, mainly related to the soil hydrological groups which, together with land use, are the most sensitive variables in runoff threshold estimation. These uncertainties are especially important in small semiarid basins and Mediterranean ephemeral streams where soils are generally very heterogeneous, not well developed, highly dependent on underlying rock and have been heavily exploited. In order to improve runoff threshold estimation in semiarid environments, some authors have included detailed information about soils in the model but without improving the results significantly. Others have assigned the soil hydrological group directly from lithology, without considering soil type. This paper proposes some modifications to the SCS method related to the assignation of the soil hydrological group in order to improve runoff threshold estimation in Mediterranean basins. In these environments, hydrological behavior is highly determined by the strong interaction between geomorphological structure, lithology and edaphology. These three variables are combined at basin scale, in order to assign the soil hydrological group with the aim of using hidrogeomorphological interpretation to improve runoff threshold estimation. The method has been applied, using GIS, to several small basins located in the Mediterranean Spanish region. Although the estimation of runoff threshold using a GIS is a process which can be easily automated, the proposed modification requires the expert geomorphological knowledge of the catchment. Runoff thresholds obtained with this method are presented here. Although the results have to be checked with runoff threshold data obtained by water budget, so far they are consistent with the expected values for the study area from literature references.