NEW LONDON, Conn. ---- In the early 1990s, sociologist Alejandro Portes and a group of colleagues began an ambitious project of following the progress of 5,200 children of foreign-born immigrants.

Immigration studies were rare then, and focused on arriving adults. Portes hit on the observation that the real national impact of the burgeoning number of immigrants turned not on inflammatory issues such as their health care, but on the success or failure of their children. In turn, there was a historical American bargain for the immigrants themselves: They might suffer, but their offspring would advance.

The Cuban-born Portes, now at Princeton, was just then developing theories that challenged that notion of upward mobility. Building on the work of the celebrated sociologist Herbert Gans, Portes and his group said that the popular perception of each generation of children doing better than their parents was wrong. Assimilation has become segmented, they said, with some children stagnating in comparison with their parents and others actually doing worse. Gans called the latter "negative assimilation."

To test his theories, Portes picked adolescents from ethnic groups in San Diego and Miami. Their average age was 14. He and his team revisited them at 17 and 24, and then went back to interview a small group who by then averaged 28. The results, dribbled out in stages, confirm his theories.

Now come his final findings, presented here last weekend during a conference at Connecticut College and published in the Journal of Ethnic and Migration Studies. And they are alarming. Most immigrant children in his sample have in fact moved up, thus overwhelmingly benefiting the country. But a sizable minority, particularly of Mexican-Americans, our single largest immigrant group, are still stagnant or have fallen into an underclass of crime, school dropouts and teenage pregnancies.

From Portes' work and other studies, we know who and why; but just as the government, community leaders, educators and parents were unable to come together in the 1950s and '60s to prevent the spiraling decay that hit African-American communities, a similar story is playing out with millions of immigrant children.

While many black Americans have since broken the spiral, and the Mexican-American situation may be less dire than it was for blacks, the numbers should be setting off a call for action.

In the Portes sample of Mexican-Americans, nearly 40 percent dropped out of high school and 20 percent of the men were in jail. Other studies show that the high school dropout rate among Mexican-Americans actually goes up in the third generation, further confounding expectations. This compares, in the Portes study, with Chinese-Americans, whose dropout rate was less than 6 percent and none were in jail. By early adulthood, there also were no teenage pregnancies among the Chinese-Americans, but 41 percent among the Mexican-Americans.

Haitians, Jamaicans and Laotian/Cambodians also fared poorly in the Portes study, but they are much smaller immigrant groups.

Measured another way, the Pew Hispanic Center reported two weeks ago that 44 percent of Latinos between the ages of 16 and 25 were in high school or college, trailing far behind the 53 percent of blacks and 58 percent of non-Hispanic whites. Nearly two-thirds of Latinos are of Mexican descent.

The reasons for the Mexican-Americans' performance are many. Their immigrant parents tend to have little education and few job skills. As happened with blacks, single-parent households and lack of parental control can lead to poor early school performance, which in turn leads to leaving school and pursuing often readily available deviant lifestyles such as gangs and drugs. Racial discrimination and low expectations by teachers add barriers.

A crucial element has been a radical shift from an industrial-based economy, with many steps up the ladder, to an information-based economy that is bifurcated, almost like an hourglass, between the educated and the unskilled. Those on the bottom have little chance to make it into the top. Previous immigrants had similar low skills and education but did not face this structural context.

Many immigrant children carry an additional weight that some social scientists say has a legacy similar to slavery: being here illegally. This status, of the children or the parents, creates deep angst that affects performance and limits opportunities.

The demand for unskilled immigrants to do low-wage manual labor will not change in the foreseeable future. Yet, as Portes concludes, the stagnation or downward assimilation of their children is "almost frightening in its implications." Americans may disagree over immigration reform, but we should all be assisting these young people who are part of us.

EDWARD SCHUMACHER-MATOS writes for the Washington Post Writers Group. Comment online at nctimes.com or contact him at edward.schumachermatos@yahoo.com.