WASHINGTON (AP) _ Congressional and administration safety investigators declared Thursday ''the evidence is persuasive'' that raising the minimum drinking age to 21 saves lives on the nation's roadways.

Citing findings by her own agency and the General Accounting Office, Diane K. Steed, the national highway traffic safety administrator, urged a House panel to stick with a 1984 law that soon will begin penalizing states that have lower drinking ages.

She expressed ''our strong agreement with Congress' intent in enacting the National Minimum Drinking Age law. It will save lives across the country, reduce younger drivers' involvement in serious and fatal crashes, and establish uniformity in place of the type of border problem that now exists between states that have not yet adopted a minimum drinking age of 21.''

Since enactment of the law, 23 states have raised their minimum drinking age to 21, bringing the total number of states in compliance to 42. Colorado, Idaho, Louisiana, Montana, Ohio, South Dakota, Tennessee, Wyoming, Puerto Rico and the District of Columbia still have lower drinking ages.

Those states and jurisdictions stand to lose 5 percent of their federal highway allotments in October unless they raise the drinking age to 21. A year later, they could be penalized an additional 10 percent of their highway funds.

Ms. Steed said one study by her agency, involving 13 states that raised their minimum drinking age between 1975 and 1982, showed that the new law ''reduced the fatal accident involvement rate among affected drivers by approximately 13 percent.''

''We estimated that about 700 lives were saved in 1984 by minimum drinking ages above 18 and that about 400 additional lives could have been saved that year by a uniform drinking age of 21,'' she told the House Public Works and Transportation subcommittee on investigations.

Eleanor Chelimsky, a General Accounting Office official, said the congressional agency had analyzed 49 separate studies by federal, state and private experts and concluded ''the evidence is persuasive that raising the minimum drinking age has had significant effects on reducing alcohol-related traffic accidents for the affected age group.''

In crashes resulting in the death of the driver, she said, the ''effects ... during differing study periods ranged from a 1 percent reduction in Massachusetts to a 35 percent reduction in New York,'' she said.

Jack DeSario, a political science professor at Case Western Reserve University in Cleveland, quarreled with those findings, saying that ''concentrating on those aged 18 to 20 is not the ultimate solution and will provide only a partial remedy'' to the problem of drunken driving.

DeSario and Peter Asch of Rutgers University both presented data suggesting that raising the minimum drinking age may accomplish little more than raising the threshhold age of inexperienced drinkers - in other words, while there is a decrease in fatal crashes involving 18-year-olds, there are more crashes involving 21-year-olds who have just begun to drink.

Ms. Steed said the Rutgers and Case Western studies were statistically flawed, but agreed that ''we are under no illusion that raising the drinking age will miraculously eliminate drunken driving'' by young people.