It is generally accepted that people with diabetes should be encouraged to abstain from smoking but there are few data on the best strategy to implement this. In a preliminary survey of our diabetic patients, knowledge of the general and specific health effects of smoking was poor. In a prospective study of 70 diabetic smokers, only 50% agreed to participate in an anti-smoking programme, and the drop-out rate was high irrespective of whether the content of the programme was general or specific for diabetes. The enrollment rate was best 2 months after the diagnosis of diabetes and the drop-out rate was highest in patients recruited immediately following diagnosis. According to self-reported data, cigarette consumption fell after the first session of the anti-smoking programme but this could not be verified by the measurement of plasma cotinine. It is concluded that an anti-smoking counselling programme based on provision of information, within the context of a specialized diabetes centre, is not cost-effective.