Some 5.7 million Americans are living with Alzheimer's, the most common form of the disease.
Oxford University studied almost 500 people, with an average age of 77 years, in 15 regions across England who were randomly given either supervised exercise and support programmes, or normal elderly care.
Nearly 500 people with dementia took part, with 329 embarking on a special exercise programme and 165 receiving their usual care.
Exercise has a reputation for helping to keep older people healthy, but a new study has found that playing video games combined with expercise - so-called "exergaming" - could help in the fight to stave off dementia.
People in the programme took part in 60-90 minute gym sessions over a four month period. They also exercised at home for one hour a week.
Surprisingly, the exercise group had the worse score for Alzheimer's disease assessment, but the difference was small - 1.4 percent - and the importance is uncertain. The exercise group were fitter, but had marginally higher Alzheimer's disease assessment scores compared with the rest. On average, those who exercised saw their score change from 23.8 to 25.2. Other (secondary) outcomes included activities of daily living, number of falls, and quality of life.
To resolve the confusion, the scientists discovered that four months of moderate to high intensity exercise training does not slow down the progress of dementia. "These benefits do not, however, translate into improvements in cognitive impairment, activities in daily living, behaviour, or health-related quality of life", they added.
"It is promising data", said Cay Anderson-Hanley, associate professor of psychology at Union College and the study's lead author.
"We don't want to alarm members of the public with dementia and their families".
People with dementia are now forced to rely on services so starved of funding that they're unable to protect them from harm and the doors of A&E, let alone provide specialist care and support.
However, the exercise group did show improvements in physical fitness. It is not much less clear whether exercise can delay the decline of those with an established dementia. We know that for Alzheimer's disease the degeneration of brain cells starts many years before symptoms start and so the likelihood of altering the disease at a late stage is less than with early intervention. Compliance with exercise was good and participants were assessed again at six and 12 months.