Soil erosion has been studied in the Viotia area, Central Greece. An integrated approach has been adopted based on portable rainfall simulator studies, on data from a permanent experimental field and on the Cs-137 technique. The ultimate aim was to develop the pedotransfer functions to construct a simple erosion model based on the Revised Universal Soil Loss Equation (RUSLE, Renard et al., 1987) in order to predict soil erosion under Mediterranean conditions on the field scale. A portable rainfall simulator similar to that described by Bowyer-Bower and Burt (1989) has been used in order to investigate soil responses to rainfall intensities of eight different soils in the Viotia area. Ponding time was recorded and runoff and sediment yield were monitored for thirty minutes after ponding, while sediment yield and runoff water were analysed for nutrient N, P, K content. An experimental erosion field with nine plots of standard 22 m length (Morgan, 1979) and slope of about 70 was established and three treatments were studied, i.e. winter wheat, fallow land and bare (herbicide treated) land. The Cs-137 isotopic technique has also been used in order to estimate long term soil erosion by defining the Cs-137 profile percentage residuals in relation to the local Cs-137 input inventory, in the grid points of a first order catchment in the Mouriki village, Thiva area, Central Greece. Rainfall simulator data showed a high variability even in the same field and with the same rainfall intensity, also between fields in runoff initiation time, in the total sediment yield and in the total runoff. This was probably due to soil complexity, scaling and spatial variability. With rainfall intensity of about 80 mm h^-1 and with a duration of 30 min after runoff initiation, it was found that the ponding time in the eight (8) fields studied varied from 1.66 to 12 minutes. The percentage of the rainwater washed away varied from 30 to 71% of the rainfall. The soil erosion rates varied from 1.05 to 3.78 t ha^-1. The nitrates in the runoff varied from 1.736 to 19.528 kg ha^-1. The total K measurements in the runoff water varied from 0.627 to 1.976 kg ha^-1 while the P concentrations from traces to 66 g ha^-1. The permanent experimental field runoff and sediment yield data during the first year of the experiment showed a high variability during the growing period. This probably depends on the rainfall pattern and intensity as recorded by the rainfall gauge and the stage of vegetation growth. Indicative measurements for a period of 43 days in early spring (March-April) 1999, with 96 mm recorded total rainfall were total runoff rates varying from 1.19 m³ ha^-1 in the fallow land to 1.4 m³ ha^-1 in the wheat and 2.37 m³ ha^-1 in the herbicide treated land. The soil erosion rates for the same period were 5.23 kg ha^-1 for the wheat, 1.89 kg ha^-1 for the fallow land and 2.77 kg ha^-1 for the herbicide treated land. Based on Cs-137 residuals the erosion and deposition rates were estimated in the grid points of the catchment area in Mouriki, Thiva area of Central Greece, using the calibration models developed by Walling and He (1997). Estimated erosion rates for the catchment grid varied from 3.54 to 95.78 t ha^-1 y^-1 while the deposition rates varied from 1.23 to 189.18 t ha^-1 y^-1. An erosion model was developed to describe and predict single event or annual soil erosion rates under Mediterranean conditions on the field scale, based on the Revised Universal Soil Loss Equation (Renard et al., 1987) for sheet erosion. This was developed with the use of SB Model Maker, software (VERSION 3.0). The model is being calibrated for local conditions and validated using sensitivity analysis facilities and rainfall simulator data. With existing soil survey data and the outputs of the model incorporated in the G.I.S. at the soil mapping unit level, the Soil Erosion Map of Viotia has been produced.