Erodibility of thirteen soils from the submountainous region of Punjab in northern India was determined by subjecting the replicated sets of selected soils packed at field measured levels of bulk densities in square soil boxes to simulated rainstorms using a micro-sprinkler based rainfall simulator. These values were compared with erodibility estimated using a nomograph, an empirical equation and a fuzzy K frequency distribution generated using the program FUZKBAS. The measured soil erodibility was significantly correlated with nomographic estimates of soil erodibility when steady state infiltration rate was used to delineate the permeability classes. The value of soil erodibility at maximum membership obtained from fuzzy K frequency distributions and the value at fuzzy centroid determined by FUZKBAS were also significantly correlated with measured soil erodibility. Estimated K values were considerably higher than the measured K values. The models and relationships used in the present study are good only as indicators of the relative susceptibility of the soils to erosion. There is a need to provide a spatial prediction and uncertainty analysis of the soil erodibility. In this regard the fuzzy logic based approach used in the present study for estimating soil erodibility appears promising as it provides freedom to choose from a range of erodibility values, each associated with a membership grade and an uncertainty assessment.