Reports of studies relating physical activity to stroke and cancer sub-types indicate inconsistent findings. Some are hampered by low statistical power, owing to a low number of events, and a failure to adjust for potential confounding variables. The purpose of this study was to relate physical activity to 12 mortality endpoints in a prospective cohort study of 11,663 men aged 40-64 years who responded to an enquiry about travel activity during a baseline medical examination conducted between 1967 and 1969. During 25 years of follow-up there were 4672 deaths. Travel activity was inversely related to mortality attributable to all-causes, coronary heart disease, respiratory disease and lung cancer, whereas the association with stroke was positive. There was evidence for attenuation of some of these associations on adjustment for potentially confounding variables. Our simplistic measure of physical activity may, in part, explain the weak associations seen.