Researchers hope to continue monitoring the participants to see if they find any delayed benefit. Scientists have speculated that nonsteroidal anti-inflammatories, such as Aleve and Celebrex, might prevent Alzheimer's by reducing inflammation in the brain or by other means.

"The drugs have several effects in the brain and the different effects could be important at different stages in the illness," said study co-author Dr. John Breitner of the University of Washington in Seattle.

Previous studies had found that people who took the drugs ran a lower risk of developing Alzheimer's. But those were observational studies, meaning they observed people's behavior and health. The people who took the pills may have had other healthy habits that lowered their risk.

The halted study included more than 2,000 people ages 70 and older with a family history of Alzheimer's but no thinking problems themselves. People were randomly assigned to take standard daily doses of either Celebrex, Aleve, also known as naproxen, or a dummy pill.

At the start and annually for up to three years, they took a battery of tests. In one, they named as many grocery items as they could in one minute.

All three groups scored about the same at the start. But over time, the Aleve takers scored on average slightly lower than the people who took placebos. The Celebrex takers scored slightly lower than the placebo takers on most, but not all, of the tests.

"There's no evidence that people should be on these drugs to prevent Alzheimer's disease," said Dr. David Bennett of Chicago's Rush University Medical Center, who was not involved in the study but does similar research. "With the side effects of these drugs, people shouldn't be taking them for this reason."

Both products now carry warnings about heart risks. Anti-inflammatory drugs also can cause serious gastrointestinal bleeding. Experts advise patients to ask their doctors about how long to take the drugs for pain.

The study was funded by the National Institute on Aging. Pfizer Inc. and Bayer Healthcare provided the drugs and matching dummy pills for the study, but did not participate in the research. Some of the authors reported receiving speaking or consulting fees from drug companies, including Pfizer.

"Unfortunately, as this paper demonstrates, research doesn't always translate into successful new treatments," said Dr. Gail Cawkwell of Pfizer in an e-mail. Cawkwell pointed out that the study did not find increased heart risks for Celebrex.

Bayer spokeswoman Tricia McKernan said the early end to the experiment reduces the relevance of the data. She said the findings don't apply to the intended use of Aleve as a short-term pain reliever.