Wash-off experiments of three Cu-based fungicides were conducted with a single raindrop simulator with known drop size and fall height. Losses were quantified as total Cu (CuT), in solution (CuL), and particulate (CuP). Cu wash-off time course was modeled for two different drop sizes using a stochastic model based on the cumulative detachment by random scattered raindrop impacts. In other set of experiments, the influence of raindrop size, fall height, and fungicide dose was analyzed statistically by means of a full factorial design. Most Cu was lost as particles sized from 0.3 to 1 μm. The stochastic model gave good estimations with two detachment performance levels. The best-fitting model parameters were as follows: the single element area covered by one impact (1.7 and 0.38 cm2 for the large and small raindrop, respectively), the average number of repeated drop impacts on one single element area that exhaust non-rainfast fungicide (4.2 ± 3.0 small drops and 2.5 ± 0.5 large drops for the high performance level, low performance level needed 30 ± 10 large drops and 40 ± 88 small drops), and the mass washed-off by a single-drop impact (from 1.27 ± 0.2 μg Cu to 3 ± 1 ng Cu per impact). Factorial design showed the dosage was the most influential factor in the three fungicide formulations. The model can help to estimate fungicide losses in field from rainfall and canopy properties. However, the particulate/soluble loss ratio cannot be predicted by the model since both the particle detachment and solubilization were not clearly related with the raindrop energy.