BACKGROUND: Some studies indicate that a large part of the beneficial effect of physical activity on mortality is confined to a threshold effect of participation. METHODS: Self-reported physical activity was investigated in relation to all-cause mortality in the Danish Diet, Cancer and Health cohort, including 29,129 women and 26,576 men aged 50-64 years at baseline 1993-1997. Using Cox proportional hazards models we investigated the associations between mortality rate and leisure time physical activity by exploring 1) participation (yes/no) in each type of activity; 2) a simple dose-response relationship with hours spent on each activity, supplemented with indicators of participation in each activity; and 3) inflexion or nonmonotonic dose-response relationships using linear splines. RESULTS: A total of 2696 women and 4044 men died through March 2010. We found lower mortality with participation in sports (for women, mortality rate ratio = 0.75, 95% confidence interval = 0.69-0.81; for men, 0.78, 0.73-0.84), cycling (for women, 0.77, 0.71-0.84; for men, 0.90, 0.84-0.96), or gardening (for women, 0.84, 0.78-0.91; for men, 0.73, 0.68-0.79) and in men participating in do-it-yourself activity (0.77, 0.71-0.84). A weak adverse dose response was seen for walking and gardening, but the association was small (1-2% increase in mortality per additional hour). We found no signs of inflexion or nonmonotonic effects of additional hours spent on each activity. CONCLUSION: Mortality was lower with participation in specific leisure time physical activities, but not with more time spent on those activities. This could suggest that avoiding a sedative lifestyle is more important than a high volume of activity. Nonparticipation in these types of physical activity may be considered as risk factors.