Environmental tobacco smoke exposure is an important health risk for small children. The development, spread and evaluation of a national child health-centre-based counselling method targeting environmental tobacco smoke is described. The work progressed in six steps. In a first step, accomplished in 1994, it was found that child health nurses used a limited repertoire of techniques and were dissatisfied with their discussion on tobacco smoke. In a second step, routine recording of parental smoking status was introduced at all child health centres. In a third step, a counselling method based on Bandura's self-efficacy concept was developed, 'smoke-free children'. In a fourth step, smoke-free children was tested by 28 nurses in 128 families. At follow-up discussions, all parents said that they now smoked outdoors and that they had cut down on their smoking. In a fifth step, the national dissemination of smoke-free children was studied. A manual and a videotape were launched in 1995, supported by a newsletter and 10 regional conferences in the following years. In January 1997, 36% of the child health nurses in Sweden (three counties excluded) stated that they used the method. Training of county instructors did not seem to have improved dissemination. In a sixth step, routinely collected information on parental smoking in Stockholm county on infants born 1995-1997 was used to study the effect. Little change in smoking rates between two consecutive years was found before the introduction of smoke-free children. Yet, after training of the child health nurses, the annual decrease was 1.7% in a pilot area and later, in remaining parts of the county, 2.7%. Thus, answers to two crucial questions were given: first, that the method seemed to affect parental behaviour; and secondly, that the training of county instructors might not have affected the dissemination of smoke-free children.