Soil erosion has been a major concern in the United States for over a century. Water erosion accounts for about two thirds of this erosion. Past researchers have divided the erosion process into rill and interrill erosion components. Rill erosion process can be further divided into scouring, headcutting and side sloughing subprocesses. Headcuts are defined as local areas of intense erosion accompanied by an abrupt elevation change. To more accurately model the rill erosion process, headcutting mechanics were developed to mathematically model the phenomenon;Headcut erosion data were collected on seven soils in six states as part of the USDA Water Erosion Prediction Project (WEPP). Rill plots 9m long and 0.5m wide, were set up on 3% to 6% slopes. These plots were formed with cultivator shovels. A rotating boom rainfall simulator was used to simulate rainfall over the erosion plots. After equilibrium runoff was achieved, additional flows of 8, 16, 24, 32, and 40 L/min were added to the top of each rill. During each flow addition, wooden stakes were placed adjacent to headcuts that formed in the rill and overhead stereo photographs were taken to determine the location of headcuts. Flow additions were repeated without rainfall. Headcut erosion was measured by calculating the volume of soil voided and multiplying this by the soil bulk density;Five models were developed using the principles of fluid mechanics. Statistical analyses indicated that two models, the total energy model and the vertical force model, were superior over other models tested. The energy model was based on the sum of the energy lost at each headcut. The vertical force model was based on the force in the plunge pool due to the flowing water. Coefficients of determination (r[superscript]2) using the energy model and the vertical force model were 0.85 and 0.84, respectively. Soil properties were correlated to the erodibilities calculated with the two models. Organic carbon had the highest correlation for the energy model erodibilities with a correlation coefficient of -0.95. Soil shear angle had the highest correlation for the vertical force model with a correlation coefficient of -0.89.