IN MARCH 2002, President Bush announced an expansion and renovation of U.S. foreign assistance. He proposed the Millennium Challenge Account, which would ultimately channel to poor countries $5 billion a year, a sum that would reflect a 50 percent expansion in existing aid programs. The new money would be spent in a new way: It would go to a short list of countries with sound policies, rather than being spread around dozens of places where corruption or political disorder undermines progress. Two years on, Mr. Bush's idea is not attracting the funding that he talked of. But his initiative has created an improved model for U.S. foreign assistance.

The new model works by measuring the quality of poor countries' policies in three areas. The first concerns governance: Do citizens have civil and political rights? How bad is corruption? Is the government effective? The second category is investment in people, particularly in immunization, other health programs and primary schooling. The third is the quality of economic policy, measured partly by indicators like the inflation rate but also by the number of days it takes to start a business and "regulatory quality." The Challenge Corp., the government entity formed to administer the new aid program, has measured 77 poor countries against these benchmarks. The top 16 have been invited to apply for grants.

Nobody pretends that these measurements are perfect; quantifying the "rule of law" is tricky. But the indicators are nonetheless the best available: They have been taken from respected outside institutions, such as the World Bank and Freedom House. Some might wonder, for example, why Armenia, a country with poor political and civil rights, was chosen, but Armenia's strong scores in economic management, rule of law and "government effectiveness" suggest that aid dollars will be well spent there. Why did Uganda and Vietnam, two development stars of the 1990s, fail to make the grade? Because their political scores were even worse than Armenia's.

The question now is how the aid will be spent in the selected countries. High-performing poor countries tend to be the darlings of donors, so finding worthwhile projects may be a challenge. One solution may be to stay away from health and education, which attract the most outside backing, and go for power, water and roads. But infrastructure projects require engineering studies, environmental assessments, social-impact evaluation -- in short, they take time to get off the ground. Another solution may be to pour money into government budgets, so that the quality of government personnel and spending is strengthened, but this approach may not appeal to Congress, which likes backing discrete projects.

Supporters of the Millennium Challenge Account worry that it is underfunded. When he laid out his plan two years ago, Mr. Bush said he wanted $1.6 billion in the current year, double that in 2005 and then $5 billion a year thereafter. Instead, Congress appropriated $1 billion for the current year, and the administration has requested only $2.5 billion for next year; the eventual $5 billion target seems distant. But the chief danger to Mr. Bush's project is that, in an effort to wring money from Congress, his administration will make unrealistic promises about how quickly aid can be disbursed and how measurable the results will be. That will only breed cynicism when the promises go unfulfilled. Fighting poverty takes patience.