BACKGROUND: The testosterone (T) responses to a physical stimulus are thought to be more stable and reproducible compared to a psychological stimulus. PURPOSE: This study compared the salivary T (Sal-T) responses to both stimuli in four groups of men: professional rugby players (n = 17), recreational rugby players (n = 10), a mixed athlete group (n = 14) and untrained controls (n = 12). METHODS: Each group completed three treatments: (1) watching a video with aggressive rugby footage, (2) performing a short bout of sprint exercise and (3) a control session. Saliva samples were taken before and 15 min after each treatment. RESULTS: The sprint exercise changes in Sal-T levels were similar in the elite rugby (17.1 ± 11.1%), recreational rugby (11.9 ± 15.9%), mixed athlete (27.6 ± 32.0%) and control groups (25.3 ± 23.6%). In response to the video, Sal-T increased in the elite rugby (6.9 ± 6.4%) and untrained groups (11.9 ± 13.5%), but decreased in the recreational rugby players (-7.5 ± 11.0%). The individual Sal-T responses to the sprints were also correlated (r = 0.69 to 0.82) with other treatment responses. CONCLUSIONS: Sprint exercise had a more consistent effect on Sal-T than a video with aggressive content and thus, could provide a reliable stimulus for increasing T availability in men with different training backgrounds. Individual Sal-T reactivity also appears to be somewhat stable across different treatments. These data provide further understanding around the induction, moderation and interpretation of T physiology.