Older people who exercise three or more times a week are less likely to develop Alzheimer's disease or other types of dementia, according to a study that adds to the evidence that staying active can help keep the mind sharp.

Researchers found that healthy people who reported exercising regularly had a 30% to 40% lower risk of dementia.

The study, published today in the Annals of Internal Medicine, didn't reach any conclusions about whether certain types of exercise helped more than others, but researchers said even light activity, such as walking, seemed to help.

"It seems like we are delaying onset," said Wayne McCormick, a University of Washington geriatrician who was one of the study's authors. "The surprising finding for us was that it actually didn't take much to have this effect."

Some researchers have theorized that exercise might reduce brain levels of amyloid, a sticky protein that clogs the brain of Alzheimer's patients.

The study, from 1994 to 2003, followed 1,740 people ages 65 and older who showed no signs of dementia at the outset. The participants' health was evaluated every two years for six years.

Out of the original pool, 1,185 people were later found to be free of dementia, 77% of whom reported exercising three or more times a week; 158 people showed signs of dementia, only 67% of whom reported they exercised that much. The rest either died or withdrew from the study.

The study couldn't determine if exercise helped prevent dementia altogether, because not all of the participants were followed up to their deaths.

The frequency of dementia was 13 per 1,000 person-years for those who said they exercised three or more times a week, compared with 19.7 per 1,000 person-years for those who reported exercising less.

Other researchers said studies -- in which participants would be randomly assigned to either exercise or maintain their usual habits -- are needed to confirm the findings.

"From now on, we should really start designing clinical trials," said Constantine Lyketsos, a Johns Hopkins University epidemiologist involved in a study that made similar findings last year. In that study, researchers tracked 3,375 people over age 65 from 1992 to 2000, surveying them on the kinds of activities they did. Those doing the widest variety of activities proved far less likely to develop dementia.