When third-grader Sharon Deleon learned in June that she had failed a standardized test trumpeted by New York Mayor Michael R. Bloomberg (R) as the key to ending "the discredited practice of social promotion," she felt upset and humiliated.

"It's not fair," she told her mother, Carmen, who is from Puerto Rico. "I'm going to be 11 years old next year, and I'll still be in third grade."

Sharon is one of more than 9,000 New York third-graders told that they are likely to be held back in September as the city begins a program to combat social promotion, the practice of advancing students to the next grade regardless of academic performance. She and others have been told that they now must demonstrate mastery of basic reading and math skills necessary to succeed in fourth grade.

A centerpiece of Bloomberg's efforts to fix the nation's largest school system, the new policy has triggered a furor in New York, with critics accusing the mayor of political grandstanding and placing too much emphasis on statistically imprecise tests as the yardstick for academic progress. It also has reopened a national debate about social promotion, a phenomenon condemned by many educators but still widely practiced.

On a more practical level, parents, teachers and school officials will be forced to cope with more than twice as many retained third-graders as the system had a year ago, if current estimates hold.

Bloomberg's supporters say the third grade represents a critical stage in a child's educational development, marking the barrier between "learning to read" and "reading to learn." They argue that children who are unable to read and do simple math by the end of third grade are likely to fall behind for the rest of their school career.

"We have thousands and thousands of [high school] students who are reading or doing math on an elementary school level," said New York City education Chancellor Joel Klein. "Almost invariably, these are the students who drop out of the system. We are trying to build the necessary skills in the early grades to prepare students for high school."

Opposition to the mayor's initiative has come from a wide cross-section of teachers' and parents' organizations, including the United Federation of Teachers, the city's largest teachers' union. Last month, the Education Committee of the New York City Council passed a resolution rejecting the new promotion policy as "ill-conceived" and based on "flawed" testing mechanisms.

Adding to the controversy is the manner in which the policy was first adopted in March, after a Gotham version of President Richard M. Nixon's "Saturday night massacre." To get the votes they needed, Bloomberg and his allies first had to fire three handpicked members of the Panel for Educational Policy, a body set up by the mayor after he took over direct control of the school system in 2002.

"It was heartbreaking," said Susana Leval, one of the dismissed panel members. "No one on the panel was in favor of social promotion, but we heard testimony suggesting that retention based on high-stakes testing is not going to work either." Over the past two decades, a growing number of school districts have pledged to end the practice of social promotion on the grounds that it contributes to low educational standards and high dropout rates. But there is much disagreement among teachers and educational researchers over the best way to combat the phenomenon, and over the role of standardized tests in determining whether a child should be promoted to the next grade.

New York launched a big campaign to eradicate social promotion in 1981, but it stalled after a few years. The effort was costly, and research suggested that students who were held back a grade derived little benefit from the experience and were more likely to drop out of school. The Chicago school district adopted a similar program in 1995, but it, too, has been scaled back.

Klein argues that the new promotion policy in New York differs from the failed 1981 program in important respects. In addition to setting stricter standards for advancement, the city has set aside $125 million for programs targeted at disadvantaged children, including summer school, smaller class sizes and personalized tutoring.

"Anybody can retain kids, that's easy," said Klein, an assistant attorney general in the Clinton administration best known for his prosecution of Microsoft Corp. on antitrust charges. "We are trying to refocus the system on why it is that students aren't learning."

Research on previous attempts to eradicate social promotion is inconclusive. Recent studies of third- and eighth-graders held back a year in Chicago suggested that they benefited little from the experience, and may have been put at greater risk of dropping out of high school.

Although some Chicago students benefited from being held back in third grade, many did not, said Jenny Nagaoka, an investigator for the Consortium on Chicago School Research and one of the authors of the third-grade study. "They don't catch up -- they just continue going on at the same pace they would have done otherwise."

Former Chicago school superintendent Paul Vallas cites overall improvement in academic performance in the nation's third largest school system as justification for the retention policy. He has introduced similar policies in Philadelphia, where he is now school superintendent. He argues that it is unfair to focus solely on retained students; a campaign against social promotion can have positive effects throughout the school system by pressuring other students to perform better.

"The retention policy played a part in getting kids to buckle down," he said. "The bottom line is that by all indicators, the school district improved during the six years I was in Chicago. Most kids did better."

In New York, critics argue that Bloomberg and Klein are paying too much attention to the results of one-hour reading and math tests administered to third-graders in April. They say that the tests were designed as tools for gathering general information about how students are performing, rather than a means of determining which students should advance to the next grade.

"There is a significant margin of error associated with these tests," said New York University education professor Robert Tobias, who helped establish the testing system as the city's director of assessment and accountability. "They can't give you a precise and absolute measure of a student's achievement."

The city has responded to such criticism by instituting an appeals procedure and giving teachers input on which students will be retained. Estimates of the number of students who will be left behind under the new policy have dwindled from more than 15,000 to fewer than 9,000, compared with about 4,000 last year. Additional modifications have been made for special-education students and non-English speakers.

The revised guidelines mean that, contrary to the draconian cutoff announced in January, thousands of third-graders who scored a failing "Level 1" on the tests still will be promoted. The final decision will be made by principals and regional school superintendents on the basis of student portfolios assembled by teachers.

"The test is like a little red flag," said Lillian Druck, principal of Public School 199 in Long Island City. "You have to look at the whole picture."

At their meeting last month, New York City Council members vied with one another to lambaste the third-grade promotion policy as unfair and unworkable. "Everything that could go wrong has gone wrong," said Education Committee Chair Eva Moskowitz. She cited reports that some students saw test answers in advance and some of the answers did not match the questions.

Others complained that the new policy discriminates against students from disadvantaged backgrounds, whose parents are unable to work the system. In some predominantly minority schools, more than half the third-graders flunked the math or reading tests or both, making them likely to be retained unless they show a dramatic improvement in summer school. Citywide, the failure rate was about 17 percent.

"If you map the third-grade test scores, you will see that it is the neediest districts that are most affected," said Robin Brown, president of the United Parents Association of New York City. "These neighborhoods get the least-experienced teachers, have the greatest numbers of health issues, the worst housing. The new promotion policy just puts a Band-Aid on the problem."

The latest concessions by Bloomberg and Klein have failed to satisfy a vocal minority of parents who have sued the city, alleging discrimination. "The kids were so stressed because of all the fuss and publicity surrounding the test that they did poorly," said Cynthia Carcagena, whose daughter, Janissa, had the lowest possible score on the reading and math tests.

Carmen Deleon attributed her daughter Sharon's poor performance in part to health problems that caused her to miss many days of schooling because of doctor appointments. "Holding her back will just cause her more emotional problems, and frustration," she said.

But at Public School 199, parent coordinator Karen DeLeon said most parents had concluded that the promotion policy was "fair" after listening to lengthy explanations about how the appeals procedure will work. "Maybe one parent was a little upset," she conceded, because her child was being urged to attend summer school, interfering with the family's vacation plans.

"The kids were so stressed because of all the fuss . . . that they did poorly," said Cynthia Carcagena, with daughter Janissa, 9. The city has an appeals procedure for those who have low scores on the one-hour reading and math exams.