Two-thirds of complaints received by the Hong Kong Environmental Protection Department in 1988 were related to poor air quality. In July 1990 legislation was implemented to reduce fuel sulphur levels. The objective of this study was to measure the impact of the intervention on respiratory health in primary school children. In all, 3521 children, mean age 9.51 years (SD = 0.78), from two districts with good and poor air quality respectively before intervention were followed yearly from 1989 to 1991. Children and parents reported the children's respiratory symptoms using self-completed questionnaires. Factor analysis was used to derive independent scores from 12 symptoms. Four groups of related symptoms were identified and binary variables (presence of any symptom in each group) were treated as dependent variables in modelling using generalized estimating equations procedures. In 1989 and 1990 an excess of respiratory symptoms was observed in the polluted compared with unpolluted district. The significant effects (odds ratio [OR], 95% confidence interval [Cl], P value) associated with living in the polluted district were: cough and sore throat (OR = 1.22, 95% Cl: 1.04–1.43, P< 0.01) and wheezing (OR = 1.35, 95% Cl: 1.10–1.66, P<0.01). After the intervention, in the polluted district only, sulphur dioxide levels fell by up to 80% and sulphate concentrations in respirable particulates by 38%. Between 1989 and 1990–1991 there was a greater decline in the polluted compared with the unpolluted district for reported symptoms of cough or sore throat, phlegm, and wheezing. The risks to respiratory health for children exposed to tobacco smoke in the home were higher than those for air pollution in both 1989 and 1990 and remained unchanged in 1991. Air quality can be improved by fuel controls but an effective intersectoral approach is required if other risks from environmental tobacco smoke are to be avoided.