Maximal oxygen uptake (VO?max) is currently the most effective indicator for assessing individual cardiopulmonary endurance, playing a significant role in formulating exercise training plans, improving health, and preventing sports injuries. Studies have shown that VO?max is significantly correlated with heart rate variability (HRV). Current research on HRV-based prediction of VO?max has not utilized frequency-domain indicators and shows limited applicability to ordinary people, necessitating further exploration of more comprehensive HRV indicator models for predicting VO?max in this population. To address this, we propose a multi-domain HRV indicator prediction method for ordinary people' VO?max based on genetic algorithm-optimized support vector regression (SVR). HRV signals were collected from ordinary college students during both resting and post-exercise recovery periods using wearable devices. The time-domain, frequency-domain, and nonlinear features of the HRV signals were extracted. Subsequently, an unoptimized support vector regression (SVR) model and an SVR model with the optimal feature combination screened by the genetic algorithm (GA), namely GA-SVR, were constructed to predict the VO2max obtained by the standard method of collection. Finally, the root mean square error (RMSE) and the mean absolute error (MAE) were applied to evaluate the performance of the models. The results indicate that during rest, the performance metrics for HRV combined with SVR were RMSE=4.440 0 and MAE=3.577 6, while GA-SVR achieved RMSE=3.976 7 and MAE=3.440 4. During recovery, the performance metrics for HRV combined with SVR were RMSE=4.280 9 and MAE=3.483 5, whereas GA-SVR attained RMSE=4.025 3 and MAE=3.288 6. The experimental results demonstrate that the proposed method achieves satisfactory performance in predicting VO?max for ordinary college students.