This study aimed to investigate whether the variable dose-response model, with estimates free to vary over time, can account for overreaching during intensified training in swimmers.A time-varying model using a recursive least squares algorithm was applied to data from eight swimmers collected over 61 weeks, comprising five training cycles. Each data set included daily training load calculated from pool kilometers and dry land training equivalents, and performance measured twice weekly from 50 m trials. Weekly changes in model parameters were used to calculate the model impulse response that is defined as the time course of performance after a single training session.Functional overreaching was evidenced by a significant decline in performance within four cycles of increased training, followed by a peak in performance after two or three weeks of reduced training. Model estimates from the time-varying model provided markers to distinguish overreaching from acute fatigue during intensified training. When an increase in training led to a decrease in performance, the characteristics of the modelled impulse responses showed a significant increase in the acute negative effect and a decrease in the delayed positive effect of a single workout.Weekly variations in estimates from a time-varying model could be useful in diagnosing overreaching from changes in the acute negative effect and delayed positive effect of training. This information provided by the model at a particular point in the training process could help practitioners to re-adjust subsequent training.