A major government study has over-estimated the future cost of obesity to the UK by at least £10bn, according to a BBC investigation.
The Foresight report put the cost to the UK by 2050 at over £45bn a year, almost half the NHS budget.
But Radio 4's The Investigation found the estimate was based on a misreading of figures from a parliamentary report.
The report's author admitted to the programme that he had made an error but claimed that it made little difference.
The calculations were based on a Commons Health Select Committee Report which estimated that in 2001, obese people cost the NHS £1bn a year.
But the calculations for the Foresight report failed to notice that figure doubled to £2bn when allowing for the costs of both obese and overweight people.
The parliamentary report also said that the overall cost to the UK economy allowing for time off work and early deaths was £7bn or 3.5 times the £2bn cost to the NHS.
Foresight looked ahead to 2050 and estimated the NHS costs would have risen to £6.5bn.
Scaling that up to find the overall cost for the UK economy they should have used a ratio of 3.5 to give an overall cost of £22.5bn. Instead they multiplied by 7 to obtain a figure of £45bn.
The man who led the team responsible for the Foresight figures was Professor Klim McPherson, an epidemiologist from Oxford University.
After being challenged over the figures, he gave a revised estimate where total costs in 2050 had shot up to £57bn.
Several days later he changed this figure, this time down to £49.9bn.
The programme sent the Foresight figures to several respected statisticians. One called them "fatuous".
Another said: "The general sloppiness of this section is evidence to me of poor quality work."
Professor David Speigelhalter, a statistician from Cambridge University, looked at the figures too.
He said the Foresight team had made "a basic logical error in their calculations".
Asked if it was embarrassing for them, he replied: "You could say that".
Using the same figures as the Foresight team, Professor Speigelhalter produced his own estimate for future costs of obesity and overweight - £34bn, over £10bn less than the original Foresight estimate.
Professor McPherson stands by his figures: "You can contest any of these multiplication factors, they are all a bit speculative but in my judgement the safest assumption was to take the most reliable costs and that is where you get the factor of seven."
His response to critics of the quality of his report was that they were wrong.
"This has been through the Department of Health, Economics Department, and has been read by experts from various of fields. Nobody has said such a thing."
Does it matter when both figures are so big?
Professor Speigelhalter says it is vital we get them right: "It shows how important it is to scrutinise any report that comes out about health issues especially making long term projections."
The Investigation - The Truth About Obesity:
Radio 4, 2000GMT, Thu 22 November.
Online from Radio 4's page.
Podcast from the website.