For those a little uncomfortable with cost/benefit analysis I'm probably about to make you very uncomfortable. Let's start with a story for context (From the EPA Newsroom):
The Bush Administration’s Clean Air Nonroad Diesel Rule...will cut emission levels from construction, agricultural and industrial diesel-powered equipment by more than 90 percent. The new rule will also remove 99 percent of the sulfur in diesel fuel by 2010, resulting in dramatic reductions in soot from all diesel engines.
When the full inventory of older nonroad engines has been replaced, the nonroad diesel program will annually prevent up to 12,000 premature deaths, one million lost work days, 15,000 heart attacks and 6,000 children's asthma-related emergency room visits.
The overall benefits of the nonroad diesel program are estimated to significantly outweigh the costs by a ratio of 40 to 1.
I'm not going to analyze this article, or try to defend any of the numbers. Since 1981, federal agencies have been required by Executive Order to perform a cost benefit analysis for all major new regulations. How can we quantify the benefits of preventing 12,000 deaths?
To value the reduction in mortality, we will calculate the Value of a Statistical Life (VSL). VSL's are not an attempt to value a particular person's life, like the courts would do in a wrongful death lawsuit. VSL's instead give an average value of a 'statistical' anonymous person.
So how do we get a VSL for the diesel rule? Suppose I asked a sample of the population the following stylized question: The government is considering a new environmental policy that will result in 12,000 fewer deaths over the next 50 years. Because the policy will result in higher transportation prices, it will raise the price of many commonly purchased products. As a result, your household expenditures will increase by some amount each year. What is the maximum amount that your household would be willing to pay in higher prices each year if this policy is put in place?
After careful analysis of the responses we might find the population is willing to absorb an average of $25 a year in higher prices for the next 50 years (an educated guess). In accordance with OMB guidelines we have to calculate the present value of these payments over 50 years. If we assume an interest rate of 5%, the payment turns out to be $180 per household in current dollars.
Currently there are approximately 111,000,0000 households in the U.S. This means a total willingness to pay of $19,980,000,000 to prevent 12,000 deaths, or $1,665,000 per premature death. Given a number like this, it's pretty easy to see how the benefits of the diesel rule exceed costs by 40 to 1.
 For the purposes of this analysis I assume that the prevented deaths are the only benefit from the rule. Obviously this is not the case, but it makes my life much easier and that's all that matters. Many of the numbers are based on educated guesses and should not be construed as accurate.
 In the Office of Management and Budget's guidelines on performing cost benefit analyses they recommend using population willingness to pay for a project as the measure of benefit.
 As any good economics student can see this question has many problems. But let's ignore those to see how we get the VSL.