Radiation absorbed by the head from hands-free mobile phone kits might be much greater than government research had found, according to a report published on Thursday by the Consumers Association. The findings will again raise concerns about possible dangers to health from the phones.

The report, in the magazine Which?, said that hands-free kits raised levels of radiation to the head from mobile phones by up to three and a half times. The Consumers Association stressed that it had done no research into whether this radiation could cause damage to the brain, but advised users to reduce the length and number of their calls.

The results contradict a study published in August by the Department of Trade and Industry, which found that hands-free kits reduced the amount of radiation reaching the head.

The Consumers Association claimed that these tests, known as Specific Absorption Tests, were flawed because they failed to take account of the changes in radiation emission across the length of the wire between earpiece and phone.

The DTI said it was examining the Consumers Association methodology and was working with Cenelec, a European body, towards establishing a definitive way of testing emissions.

The Federation of the Electronics Industry, the trade association that represents mobile phone manufacturers, said it was reviewing the Consumer Association study, but noted that the Consumer Associations of Australia and New Zealand had carried out similar tests and found no increase in emissions from hands-free kits.

* Mobile phones are providing an unexpected health benefit, according to a paper in Friday's British Medical Journal, Clive Cookson writes. They are driving down the rates of teenage smoking.

Clive Bates, director of the pressure group Action on Smoking and Health, and Anne Charlton, a professor at Manchester University in north-west England, say the dramatic rise in teenagers' use of mobile phones - from a very low level in 1996 to more than 70 per cent today - is associated with a sharp fall in smoking among schoolchildren.

The authors cannot prove a causal link but they say that recent social attitude surveys provide strong evidence.