Does exercise reduce all-cancer death rates?

A reanalysis is made of earlier data relating to initial physical fitness and the likelihood of death from all forms of cancer. It is argued that the original analysis may have been biased by an association between initial fitness and other health habits, particularly cigarette smoking. The association with fitness status remains after reanalysis of the data on the assumption that current smoking leads to a uniform doubling of the risk of cancer death, but the effect is weaker than previously reported. There remains some potential bias, in that the quantity of current smoking may have been linked to fitness status. Some 55% of deaths were untraced, but it is argued that any socioeconomic or other bias from this cause is likely to account for the association between cancer risk and low fitness status. Any reduction of cancer risk is associated with the change from an extremely sedentary to a moderately sedentary lifestyle. It thus cannot be explained in terms of the mechanisms previously invoked to explain low risks of colonic and reproductive cancers in endurance athletes.