If you're trying to give up smoking the best way to do it is to go "cold turkey", a new study has found.
Scientists from Oxford University found that smokers who try to kick their habit bit by bit are less likely to succeed than those who decide to give it up in one go.
The team recruited 697 smokers who had chosen to stop smoking for the study and split them into two groups.
The first group - the abrupt cessation group - stopped smoking completely from day one. The second group - the gradual cessation group - set a quit day but gradually reduced their tobacco use in the two weeks leading up to that date.
According to the Press Association, Dr Nicola Lindson-Hawley, who led the British Heart Foundation-funded study, said: "Both groups had advice and support and access to nicotine patches and nicotine replacement therapy, like nicotine gum or mouth spray."
One month after their designated quit day, 49% of the abrupt cessation group were still not smoking compared with 39% of the gradual cessation group.
The scientists therefore concluded that the cold turkey participants were 25% more likely to have been successful than the group who attempted to quit gradually.
Dr Lindson-Hawley said: "The difference in quit attempts seemed to arise because people struggled to cut down. It provided them with an extra thing to do, which may have put them off quitting altogether. If people actually made a quit attempt then the success rate was equal across groups.
"We also found that more people preferred the idea of quitting gradually than abruptly, however regardless of what they thought they were still more likely to quit in the abrupt group."
But she added that for smokers who find it impossible to make a clean break from tobacco, it was better to try to cut down than simply give up.
"For these people it is much better to attempt to cut down their smoking than do nothing at all and we should increase support for gradual cessation to increase their chances of succeeding," she said.
The research is published in full in the journal Annals of Internal Medicine.