>"Cigarette smoking is the single most preventable cause of
>premature death in the United States. Each year, more than
>400,000 Americans die from cigarette smoking. In fact, one in
>every five deaths in the United States is smoking related."
>
>Well, that's amazing given that 25% of people in America
>smoke. So apparently if you die and you're a smoker, there's
>near 100% chance smoking is what killed you. Not a heart
>attack you would've had anyway, etc.

Let's assume that your both figures are correctly. 20% of the smoking deaths were due to smoking and 25% of Americans smoke. Let's count the corpses. S for smoker deaths, D is for non-smoker deaths. Let us assume that there are X americans, who without smoking have a chance of death that is Y. Let us assume that smoking increases the amount of deaths with a multiplier Z. T means total deaths. We have a statement, which claims that 20 % of American deaths are smoking-related. This means that

T = XY + F = 1.25*XY,

where F is deaths directly caused by smoking. In the study, we're going to have target groups that consist of smokers and non-smokers. For smoker deaths, we get:

S = 0.25*XYZ

For non-smoker deaths, we get

D = 0.75*XY

For total deaths, we get:

T = D + S = 1.25*XY

By placing D and S, we get

0.25*XYZ + 0.75*XY = 1.25*XY

By dividing both sides with XY and subtracting 0.75 from both sides, we get

0.25*Z = 0.5 => Z = 2.

This means that the risk of death as a smoker is twice the normal risk. Thus, 50% of the smoker deaths are smoking-related.