The kind of exercise that proves beneficial involves activities such as brisk walking, cycling and gardening.

However, less energetic activities are of little benefit, the researchers found.

11,000 graduates of Harvard University took part in a study that started in 1962.

Brisk walks

Researchers followed their progress and found that those who did the equivalent of one hour's brisk walking five days a week were half as likely to suffer a stroke.

Half as much effort produced half as much benefit - the equivalent of 30 minute's brisk walking five days a week cut the risk of stroke by a quarter.

Dr I-Min Lee, of the Harvard School of Public Health, published his team's results in the Stroke medical journal.

Exercise could help prevent stroke in the elderly

"Not only did we find that physical activity is associated with a reduced risk of stroke, but we also have some ideas as to how much and what type of activity might work best."

More intense exercise could cut the risk even further, he said.

"We found that doubling the effort showed an even greater reduction in stroke risk."

Long-term study

The people studied in Dr Lee's research joined the study in 1962. Fifteen years later they were questioned about how much exercise they did.

Two decades later, the researchers looked at how many of the graduates had a stroke. Those who exercised more were less likely to have suffered one.

The doctors took account of other factors such as smoking, alcohol consumption, blood pressure and diabetes, which are all known to affect stroke risk, and found that exercise had a strong protective effect against stroke.

Dr Lee said: "Walking, stair-climbing, and participating in moderately intense activities such as dancing, bicycling and gardening were shown to reduce the risk.

"Light activity such as bowling and general housekeeping activity did not have the same effect."

Stroke in the UK

Every year, 100,000 people in the UK suffer strokes for the first time.

The Stroke Association said that the research was part of a growing body of evidence confirming the benefits of exercise.

Professor Shah Ibrahim is a member of the association and a professor of epidemiology at Bristol University's department of social medicine.

He said the study confirmed the findings of other research produced in recent years, although the results were not representative of average members of the public as the subjects were all Harvard graduates.

He said the most important result of the research was the public health message it sent out.

"The message that vigorous exercise is not necessarily what is required but increases in daily habitual activities is useful."

He added that doctors would need to conduct further trials to establish if exercise was the real reason for the reduction in stroke.

Growing evidence

The research is the second study published in the space of a month to provide firm evidence that exercise really is good for you.

The British Heart Foundation said that the finding supported their recommendations for exercise.

"We recommend you need to do 30 minutes a day, preferably every day of the week but at least five days a week," a spokesman said.

He said that the half hour of activity could be split into three 10-minute segments through the day.

He said the kind of activity highlighted in the research was the right type.

"We tend to use the phrase physical activity rather than exercise because people tend to associate the term exercise with jogging, football, or playing a sport of some kind."

Exercise scheme

The foundation is developing a health walks programme in association with the countryside commission to help people realise the benefits of regular activity.

At present there are three pilot schemes in operation but the foundation hopes the scheme will go nationwide in the future.

Pat Knight is health walks co-ordinator in Sonning Common, Berkshire.

She said that as evidence of the long-term benefits of exercise grew, doctors felt more comfortable prescribing exercise.

The scheme was a particular success because it offered local access to regular activity.