f the lingering question about the economic boom of the late 1990's remains how widely or narrowly were the winnings shared, the answer has long been expected to lie in the 2000 census data on income and poverty.

But as the numbers began rolling out in recent weeks, they struck some people as counterintuitive: Amid the longest peacetime expansion ever, median income actually dropped and poverty rose in a handful of surprising places.

Now demographers and economists are using those data and others to piece together a picture of who made a killing, who merely got ahead and who fell behind, and to figure out why a few places widely assumed to have flourished now look as though they did not.

These things are certain. Nationwide, prosperity increased during the decade; incomes rose and poverty in most places declined. And, economists and sociologists say, the gains landed disproportionately in the laps of a small fraction of Americans already exceptionally well off.

Inequality increased, as it has since the late 1970's, but at a slower rate nationally than in the 1980's, some experts said. Other data suggest, according to one economist, that the percentage of families earning less than $25,000 dropped significantly in the second half of the 1990's.

The fate of people in the middle seems less clear-cut. While the median household income rose in the Northwest, South, Midwest and Southwest, it dropped in parts of the New York metropolitan area, in some other Northeastern cities and in swaths of Southern California.

Many experts suspect immigration helps explain those patterns. As low-wage workers poured in from places like Central America in pursuit of a better life, the midpoint on the income spectrum in some places with large foreign-born populations may have, paradoxically, inched down. At the same time, native-born people moved out — from the New York region to the Southeast, from Los Angeles to places like Nevada, some demographers say. Though those people may have moved up economically by moving away, their departures may have helped lower the median in the places they left behind.

"You see how the arithmetic works?" asked Gary Burtless, an economist at the Brookings Institution. "When you have a lot of people entering from the rest of the world, and many of them enter at the lower rungs of the wage distribution, then you can have a situation where everyone is prospering and the median income is declining."

But immigration is not the only possible explanation. Americans may tend to forget that the decade began with a recession; incomes for many people declined in the early 1990's and poverty increased. In some places, like the New York region, the recovery did not begin until the second half of the decade. Only late in the 1990's did median income catch up and exceed the 1989 levels.

"Between 1989 and 1993, there's a big drop in median income," said Edward N. Wolff, a professor of economics at New York University who specializes in income and wealth inequality. "It wasn't until 1998 that things got back to where they were in 1989. So by 2000, median income was a bit ahead of where it was 10 years ago, but not much."

At the same time, the job base shifted in some places; lower-wage jobs replaced higher-wage ones. When defense spending shrank and aerospace-industry jobs in Southern California dwindled, what took root were jobs in the garment industry and service sector, said Bruce Katz, director of the Center on Urban and Metropolitan Policy at Brookings.

"All of us here have been trying to figure out what do these regional variations mean," Mr. Katz said. "How much of it is attributable to the demographic shifts that are happening in these metropolitan areas and how much is attributable to the underlying economy?"

Then there are those who have doubts about the data itself.

Eric Kober, director of housing, economic and infrastructure planning for the New York City Department of Planning, described as "extremely counterintuitive" the census's finding that the percentage of people in the civilian labor force and the percentage of people employed actually declined nationally and especially in New York City.

Mr. Kober said that the census data for the 25 biggest cities showed that the percentage of people employed tended to drop most in cities with large numbers of foreign-born residents while the rate increased in affluent cities with lower percentages of foreign-born people, like Denver and Columbus, Ohio.

Mr. Kober suggests that the numbers are wrong. He traces the problem to what he says is a growing unwillingness by people in certain groups to answer the detailed income questions on the census long form. He contends that the data underestimate employment and income in places like Los Angeles and New York City, where median income dropped in four boroughs.

"They may be skewing the national data and accounting for this situation we're seeing, where there is a 50 percent decline in welfare nationally and yet a decline in the percentage of adults working," he said. "It can't be."

On average, median household income rose by nearly $3,000 nationally between the 1990 and 2000 censuses. The states with the biggest increases ranged from Colorado and Oregon to Mississippi. Those with the smallest increases included California, New York, New Jersey, Rhode Island and the District of Columbia. In Connecticut, on average, median household income dropped.

William H. Frey, a demographer at the University of Michigan, sorts the states roughly into three types: "Mature melting-pot states" with large number of foreign-born who make relatively low wages, like New York and California; "emerging growth states" in the South and West, growing through domestic migration; and "heartland states" attracting few immigrants and losing domestic migrants.

In the melting-pot states like California, Professor Frey sees a "barbell economy" with growth at the bottom and the top and a thinning out for the group in the middle. In the emerging growth states, like Colorado, incomes were strongly up. In the heartland states, incomes rose moderately as the aging baby boomers who dominate those places moved into their high-earning years.

Even within some states and metropolitan areas, there is considerable variation. In California, median income declines were heavily concentrated in the Los Angeles region, in counties including Los Angeles, Orange, Riverside and San Bernardino. In sharp contrast, there were big increases in median income in counties in the San Francisco Bay area and around Silicon Valley.

In many places with big increases, the gains went disproportionately to people in the top fifth of the income spectrum, according to Andrew A. Beveridge, a sociologist at Queens College who analyzed the census data for The New York Times. People at the top gained the most and people at the middle and bottom gained considerably less.

In San Francisco County, for example, where the median household income rose by $11,854, to $55,221, the income of people in the top fifth of the income spectrum rose by $33,584, to $117,730, Professor Beveridge found. The income of the people in the bottom fifth rose by $3,374, to $21,153. The income of people in the middle rose by $16,961, to $70,470.

"Most of the prosperity has gone into the hands of the rich," said Professor Wolff of N.Y.U., citing his own findings from other data. "The top 20 percent did well in the 90's. And even among that group, it's the top 1 percent that made out like bandits."

He said the pattern was especially pronounced in New York state. Citing city and state statistics, he ascribes the pattern to two things: the high concentration of New Yorkers working in industries like finance that did especially well in the 1990's, and the abundance of others working in low-wage service-sector jobs catering to the first group's needs.

"For every investment banker, you have one or two delivery men to feed them during their long working hours," Professor Wolff said. "So, in a sense, the high-income group in the city also creates this low-income group. It's a sort of symbiotic relationship."

Though it is not possible to draw conclusions about economic and social mobility based on the census data released so far, Jared Bernstein, a senior economist at the Economic Policy Institute, a liberal research organization in Washington, said other data he had studied suggested that the rate of movement out of the lowest income categories picked up in the late 1990's.

Comparing the 1980's with the 1990's, he said, "These were both eras wherein economic growth was not shared equally. But in the 1990's, the latter half especially, families were more prosperous throughout the income scale. It didn't reverse the tide of inequality. It just slowed it down, to some degree."