If this is the summer of tax cuts and school reform in Washington, 2002 is likely to be the summer of welfare revisited. The nation's bold experiment in transforming the welfare system expires next year, and Congress will need to make some difficult choices in deciding on its next phase. It will need to learn some important lessons from the experiences of people such as Sharon Suggs and Patricia Burgess.

Suggs is a 22-year-old single mother of two in Milwaukee. After dropping out of high school, she went on welfare. But under Wisconsin's W-2 reform program, which took effect in 1995 and served as a model for the federal law, she faced a 60-month lifetime limit on benefits. She received grants based on her job status and was given help in finding work.

And she found a good job, as an office assistant, earning $360 a week, more than double her welfare income. That was in the spring of 2000. But as the Milwaukee Journal-Sentinel reported, she was fired after three months, and hasn't been able to land a new job since. She complains that the only jobs available to her are low-paying, and that it's hard to reach the caseworkers at the nongovernmental agency charged with helping her.

For Patricia Burgess, the story has been very different. Also a single mother of two in Milwaukee, Burgess received clerical training to prepare her for work. As she trained, the W-2 program provided day care for her kids. When her landlord threatened to evict her, W-2 paid her rent. She got a job last December as a medical receptionist, at $9 per hour, and she's still there.

There's no disputing the fact that Wisconsin's program has produced dramatic reductions in the welfare caseload. In the three years ending last September, the number of people on cash assistance dropped by half. The real question is how long they are able to stay off welfare.

A new report by Wisconsin's Legislative Audit Bureau, an arm of the state legislature, provides good reason to be cautious. The audit bureau examined the history of 2,129 people who left W-2 in the first three months of 1998. One-third of them had incomes above the poverty level when benefits from the earned income tax credit were included. One-third of them had incomes below the poverty level. (The other third filed no returns, and thus couldn't be evaluated, but it's likely most were below the line as well.) Few of those who left the program could keep themselves out of poverty without supplemental governmental help. Barely one-fifth of the people who left W-2 in early 1998 had incomes above the poverty line without the earned- income credit.

Keeping people off welfare often proved daunting. More than a quarter of those who left the program had returned for cash or other services by July of last year. In Milwaukee, which has the lion's share of Wisconsin's welfare caseload, as well as many of its most difficult cases, 42 percent of W-2 clients were back in the program because their jobs hadn't worked out.

The bottom line seems to be that W-2 has been highly successful in getting people off welfare, less successful in keeping them off, and even less successful in lifting them into an economically secure mode of life. The W-2 alumni with the highest incomes are those who have received extensive job training and other support services and then moved on to unsubsidized private employment. Those squeezed off the rolls into subsidized jobs don't fare nearly as well.

So we know that the first stage of welfare reform has helped nudge those easiest to train and place into new jobs. We also know that it gets harder from there. The challenge for Phase II is dealing with clients such as Sharon Suggs, who find work but can't keep it, and then can't find anything that pays as well as the job they lost. And, of course, there are those who are even worse off--welfare recipients who never made the jump to work at all, and who now are nearing their 60-month lifetime limit. As the economy softens, caseworkers worry that starter jobs might dry up.

All the evidence suggests that moving people out of poverty on a permanent basis will require better "case management": pulling together job training, child care, housing and food support, health care and the other services that families need to make the jump from welfare to work.

That, in turn, requires good management in general. Critics in Wisconsin have complained about the poor track record of some nongovernmental W-2 contractors, and about the large profits they earned from the big decline in the caseload. The audit bureau study suggests that the state's welfare agency might well have to take a strong hand on the reins to improve local operations.

Creating more success stories like that of Patricia Burgess is the central problem of welfare reform's renewal. With the economy softening and the easy job placements already made, her story helps define what real reform means: weaving together a tight and broad support net, and tailoring its dimensions to fit each individual participant.