This paper proposed an approach to predict the breakthrough curves for herbal medicine adsorption. The saponins adsorption process was taken as an example. The Yoon-Nelson model was used to fit the breakthrough curves of saponins and the model parameters in different batches were obtained. The model parameters were correlated with saponin concentrations in loading solution and flow rate using polynomial regression models. According to the polynomial regression models, the breakthrough curves could be predicted and the optimal adsorption end-point could be calculated early-on. This research provided a new method for predicting the end-points of the adsorption process of herbal extracts.