NEW YORK (AP) _ Nearly two-thirds of this year's National Merit Scholarship semifinalists are boys, according to a report Tuesday that blamed sex bias in the standardized test the awards are based on.

An estimated 62.9 percent of the semifinalists in 1989 are male, according to the third annual analysis of the national awards program conducted by FairTest, a Cambridge, Mass.-based research group that has frequently criticized standardized testing.

Thirty-one percent of this year's semifinalists are girls, according to the state-by-state analysis of first names. The genders of the remaining 6 percent couldn't be determined from their names.

The group reached similar findings in two previous surveys. In 1987-88, it found 60.1 percent of merit scholars were boys, and in 1986-87, 61 percent were boys.

In all, there are 15,000 high school students who are semifinalists in the competition for about 6,000 National Merit scholarships.

FairTest has argued that the National Merit Scholarship Corp., based in Evanston, Ill., unfairly distributes its awards because it relies solely on the results of the Preliminary Scholastic Aptitude Test to determine semifinalists. The PSAT is a multiple-choice test similar to the Scholastic Aptitute Test, which critics accuse of sex bias.

Marianne Roderick, the corporation's executive vice president, disputed the bias charge in a telephone interview: ''Of course, we wouldn't use a test that's biased. We have been assured by the College Board and the Educational Testing Service that the test is not biased.''

The College Board sponsors the PSAT, taken annually by more than 1 million students, mostly high school juniors; and the SAT, taken by about 1.5 million students a year. The Educational Testing Service administers both exams.

Test critics claimed a key court victory in Februarywhen a federal judge ruled that New York State scholarship programs discriminated against women because they relied exclusively on SAT scores.

''There's no excuse for National Merit's continued reliance on a biased test when a federal court has ruled that using similar exams to select scholarship winners constitutes sex discrimination,'' said Cinthia Schuman, FairTest's executive director.

The ETS an College Board have repeatedly argued that the court decision did not brand the SAT as biased, but merely ruled that the test should not be taken as the sole criterion for making state scholarship decisions.

Roderick said that while the PSAT is used to determine semifinalists, actual scholarship winners must produce an array of information about their academic record, leadership potential and other personal qualities.

''In 34 years, we have never based our scholarship decisions solely on a test,'' she said.

She said it was too soon to confirm FairTest's figures for 1989 since scholarship winners won't be announced until later this month. In 1988, she said, 62 percent of all merit scholars were men.

However, in the ''Achievement Program'' aimed at awarding 700 merit scholarships to black students, women outnumbered male winners by 58 percent to 42 percent, said Roderick. That competition requires information about high school coursework, leadership qualities and other criteria.

The non-profit corporation distributes some $24 million merit scholarships a year, with individual awards ranging from $500 to $8,000. Some 6,000 high- achieving students win the scholarships annually.

The dispute over whether standardized tests like the SAT and PSAT are biased against female students has heated up lately.

Following the New York State court victory by the American Civil Liberties Union, a report last month by the Center for Women Policy Studies identified 23 test questions used in a 1987 SAT exam in which one sex outperformed the other by at least 10 percent. In all but two of those questions, boys bested girls.

The College Board denies the charge of test bias, and argues instead that women are less likely than men to take college prep courses in high school.