In athletes, exercise training induces autonomic nervous system (ANS) adaptations that could be used to monitor training status. However, the relationship between training and ANS in athletes has been investigated without regard for individual training loads. We tested the hypothesis that in long-distance athletes, changes in ANS parameters are dose-response related to individual volume/intensity training load and could predict athletic performance. A spectral analysis of heart rate (HR), systolic arterial pressure variability, and baroreflex sensitivity by the sequences technique was investigated in eight recreational athletes during a 6-mo training period culminating with a marathon. Individualized training load responses were monitored by a modified training impulse (TRIMP_i) method, which was determined in each athlete using the individual HR and lactate profiling determined during a treadmill test. Monthly TRIMP_i steadily increased during the training period. All the ANS parameters were significantly and very highly correlated to the dose of exercise with a second-order regression model ( r² ranged from 0.90 to 0.99; P < 0.001). Variance, high-frequency oscillations of HR variability (HRV), and baroreflex sensitivity resembled a bell-shaped curve with a minimum at the highest TRIMP_i, whereas low-frequency oscillations of HR and systolic arterial pressure variability and the low frequency (LF)-to-high frequency ratio resembled an U-shaped curve with a maximum at the highest TRIMP_i. The LF component of HRV assessed at the last recording session was significantly and inversely correlated to the time needed to complete the nearing marathon. These results suggest that in recreational athletes, ANS adaptations to exercise training are dose related on an individual basis, showing a progressive shift toward a sympathetic predominance, and that LF oscillations in HRV at peak training load could predict athletic achievement in this athlete population.