The news, published in this week's Journal of the American Medical Association, was the latest in a sometimes confusing stream of reports on estrogen combined with progestin to combat post-menopausal problems.

The hormones have been shown to halt or reverse osteoporosis, lessen the risk of hip fractures and prevent uterine cancer. But a major government study on long-term use was halted last summer after it showed the estrogen-progestin combination sold as Wyeth's Prempro carried an increased risk of ovarian cancer, heart attack and stroke.

Studies are continuing to determine whether estrogen alone is safer when not combined with progestin; drug makers also believe lower doses of the drugs are safer.

Tuesday's report from Wake Forest University Baptist Medical Center involved an offshoot of the study which was halted.

It found the hormone combination doubled the risk for probable dementia in women 65 and older and did not prevent mild cognitive impairment. Translated to a population of 10,000 women taking the combined therapy, that would mean an additional 23 cases of dementia per year, it said.

The study involved 4,532 women, about half of whom received the drugs while the rest were given an inert placebo.

Wyeth issued a statement saying it had revised its labeling to reflect the findings, but also said the average age of women in the study was 71, a level where the risk of dementia is higher to begin with than 51, the average age when therapy starts.

"The overall individual risk to women is low, although there is reason for concern," said principal investigator Sally Shumaker. "Because of the potential harm and lack of benefit found, we recommend that older post-menopausal women not take the combination hormone therapy to prevent dementia."

The Food and Drug Administration issued a statement saying it will not require any changes in labeling for affected products but in general women should talk to their doctors before using estrogen-progestin products. They should "use the lowest dose for the shortest duration to reach treatment goals, although it is not known at what dose there may be less risk of serious side effects," it said."

The same researchers, in a separate report, also looked at overall cognitive functioning in the women, including concentration, language, memory and abstract reasoning.

They found that women taking estrogen-progestin performed slightly worse than the placebo group.

The study was funded by Wyeth Pharmaceuticals and the National Institutes of Health.

Wyeth has won U.S. government approval to market lower-dose forms of estrogen-only Premarin and Prempro. Since the government test on the stronger strength drug was halted last year, combined sales of Premarin and Prempro have plunged 40 percent.