> Exercising (moderate physical activity) for 2-2.5 hours per week leads to a 20% reduction in all-cause mortality (not to mention its many other benefits). Exercising for an additional 5.5 hours leads to another 20% reduction.

With a very important caveat in the report:

> Because all of the studies in the evidence base were observational epidemiologic studies with no randomized controlled trials, the data cannot prove causality of effect.

People who don't feel that great when they exercise (despite everyone urging them to exercise) are probably not in the best health. The pervasive advice to exercise more will by itself confound the studies, because healthier people will tend to self-select into the "exercise" group.