AMID MUCH partisan mudslinging, Congress failed before the election to finish one of its most important tasks: the reauthorization of the 1996 welfare reform legislation, which stalled when the House passed a flawed bill and the Senate failed to pass any bill at all. In January, a Republican Congress and administration will have a chance to revisit the debate. Instead of returning to the faulty House reauthorization bill, all involved should look again at what the original welfare reform legislation actually did and did not achieve before deciding how -- and if -- it needs to be changed.

On the achievement side, the list is long. Five years down the line, millions of families have left welfare for work, welfare caseloads are at historical lows, child poverty has declined by some measures and the number of children born out of wedlock has stopped rising. One example of unexpected success: In the District caseloads have dropped by a third. Of the 16,000 who remain on the welfare rolls in the District -- which is known for a high proportion of difficult cases -- 11,000 are engaged in some kind of work or training. Contrary to expectations, the economic slowdown of the past year has not substantially reversed these gains, in the District or elsewhere. Nor have any studies shown that the children of working single mothers have suffered in any measurable way. Cause-and-effect is notoriously difficult to measure, but the reform legislation surely has played a major role in these gains.

On the other side of the balancing scales, there is still no evidence that those who did leave the welfare rolls, whether working or not, have managed to raise their incomes. The average ex-welfare recipient earns about $8 an hour and isn't moving up. Some of those who initially got jobs also proved unable to keep them, for a variety of reasons: lack of transportation, lack of education, drug addiction, depression. A few returned to welfare -- about 25 percent of those who initially got jobs in the District. Some depend on parents or friends; others survive in the cash economy. Aware that some people are in jobs but barely hanging on, or else out of the system altogether, many states have used the money saved on lower welfare bills to pay for child care, transportation and education and training. In the case of the District, where illiteracy is a major problem, surplus federal funds have been used to help women pass high school equivalency tests, among other things, as well as to subsidize both formal and informal child care. While welfare rolls are substantially down, most states still do not want to eliminate these kinds of services to low-income people and believe it will take more than five years to alter the dependency culture for good.

What we are looking at, in other words, is probably best described as an experiment, largely successful thus far but unfinished. We also are looking at a debate that has changed radically. Among those who study poverty and administer programs, there is now general agreement on the importance of work and marriage, as well as a consensus that the states are in the best position to determine how to promote them. Federal funding, indexed for inflation, should remain in place, not least to continue providing child care at a time when state budgets are being cut. There is no need yet for the federal government to raise the number of hours women should work, nor to dictate how many hours they should work as opposed to how many they should study, nor to score political points on those grounds. The states already have plenty of incentive to get single mothers into jobs quickly, and plenty of incentive to get them earning enough money to support their families. The strength of the original bill lay in its flexibility, and that is what the reauthorization should try to preserve.