A pool of two Italian rural population samples made up of 1,712 men ag
ed 40-59 at entry was studied in 1960 and than followed up for 25 year
s. The multivariate analysis of the first major coronary event using t
he Cox model showed, in men aged 40-59, 45-64 and 50-69, the significa
nt predictive role of age, systolic blood pressure, serum cholesterol
and cigarette smoking, but not of body mass index, without marked diff
erences attributable to the aging process. Changes in systolic blood p
ressure, serum cholesterol and cigarette smoking occurring between yea
r 0, 5 and 10 of follow-up, as defined by two different indicators, in
creased significantly the predictability of coronary events occurring
between years 10 and 2 5 of follow-up when added to the model includin
g the baseline factors. It is inferred that increases and decreases (e
ven of relative nature) of the three major risk factors around the ent
ry levels are associated with higher and lower levels of coronary risk
in the 15 years after the changes have occurred.