Abstract: The research discussed the issue from three aspects, the saliva components, saliva collection and detection methods, as well as the effects of training on saliva biomarkers. The results has found that the content of most biomarkers in saliva is about 1/10 to 1/1000 of that in blood, and the lower content requires higher accuracy of saliva detection methods. Besides, passive drooling is a commonly used sampling method, while the time point of saliva collection and its analysis method have not been standardized. Salivary testosterone, cortisol, uric acid and other biomarkers have strong correlations with the content in the blood, which provides evidence for the application of saliva samples. It takes time for testosterone and cortisol to transfer from the blood to saliva, so it is recommended to collect saliva samples for about 10-20 minutes after a short-time high-intensity training to detect the content of saliva testosterone and saliva cortisol. Salivary cortisol increases and testosterone decreases after long-term high-intensity exercises while salivary α-amylase increases significantly after aerobic exercises. In the future, a lot of research is to be required to explore the relationship between different sports training and saliva biomarkers; saliva samples are to be established in the use of collection, detection, analysis and evaluation in sports training.