Our average age at death soars in the last third of life, and the longer you live, the longer you’re likely to live. Why does the total number of years we’re expected to live increase with age? Survival of the fittest. “As you work your way through the age range, you’re essentially weeding out the frail from the population,” explains Bob Anderson, chief of mortality statistics at the Centers for Disease Control and Prevention. “When you get to the older ages, you’re left with the more robust in the population. And that continues as you move up the age range.”

Death registration is mandatory in the U.S., and the CDC compiles demographic information from death certificate data filed by state vital records offices. That means we know the age of almost every person who dies in a given year — minus a few people whose bodies are not found until later — and why they died. Life expectancy is usually discussed as the number of years a hypothetical infant born in a certain year could be expected to live, and it’s based on overall mortality statistics for that period. But the average age that, say, a 65-year-old alive in that year (2007 in our graph below) will reach will be better than that of anyone younger. That’s because the older person has cleared more of life’s deadly hurdles, Anderson says.

The first hump happens in the first year of life, when around 30,000 babies succumb to problems like congenital defects, prematurity and SIDS. “Once you get past that first year, then your probability of survival is quite a bit higher,” Anderson says. Then come the teens and twenties, when accidents and violence peak. For men, especially, these are the risk-taking years, and ages 20-24 are known as the “accident hump.” In 2007, more than 10,000 men in this age group were killed in an accident or assault. (Men’s tendency to take risks is one explanation suggested for why women live longer. The protective effect of estrogen on a woman’s heart is another. Estrogen declines after menopause, which may be one factor in why the life expectancy gap between the sexes narrows from five years at birth to a couple months at age 100.) Risk-aversion begins to kick in around 25, and natural causes start to rank. By 45, cancer and heart disease are the leading killers of men and women, not accidents. (Suicides, incidentally, are also most common among 45- to 49-year-olds.) Once the frail first year, the risk-taking young-adult years and the disease-prone middle age have passed, the average age a person will achieve rises dramatically. Whereas a baby girl in 2007 could be expected to reach 80, an 80-year-old woman’s expected age of death was 89 that year.

Longevity in a large, aging population comes with a cost: proportionally more elderly people for every younger working person. “That means from a per capita standpoint, it’s going be more expensive to maintain health care and social security,” Anderson says. “Just from a demographic standpoint, it’s clear to me that it makes things more expensive.”

A shorter version of this story appeared in the July 12, 2012 issue of Fortune.