Reliable mechanical weed control requires knowledge of the achievable levels of weed control and crop damage when using certain implements in specific conditions. Quantitative methods that use weed, crop, soil, and cultivator characteristics to predict weed control and crop damage need to be developed. To that end, the relative susceptibility of weeds and crop plants to mechanical weeding and the selective ability of cultivators need to be quantified separately. The method presented in this study uses measured plant anchorage forces to quantify crop and weed sensitivity to being uprooted by a weed harrow and predicts the relationship between weed and crop uprooting by mechanical weeding. Uprooting and anchorage force of young perennial ryegrass and garden cress plants were measured in laboratory harrowing experiments on sandy soil. A nonlinear equation was introduced to describe the relationship between weed uprooting and crop uprooting. The parameters representing the selective potential of the actual crop–weed condition (K pot) and the implement selective ability (K cult) did not depend on crop uprooting. The relationship between potential weed and crop uprooting that could theoretically be obtained by a perfectly selective implement (i.e., pulling each plant with equal force) was calculated from plant anchorage force distributions measured before harrowing. The observed uprooting percentages achieved by harrowing were lower than the potential uprooting percentages. With K cult accounting for imperfect weeder selective ability, prediction accuracy was satisfactory. Field validation is required to confirm whether this method improves comparison and prediction of weeding performance of different weeding implements in different crop–weed situations.