OBJECTIVE: Data from two consecutive household surveys undertaken in mid-80s and mid-90s allow to characterize and analyse secular trends in infant and child respiratory diseases in the city of S. Paulo, Brazil. METHODS: The two surveys included random population samples aged from zero to 59 months (1,016 in the period of 1984-85 and 1,280 in 1995-96). In both surveys the instant prevalence of high respiratory diseases (above the epiglottis) and low respiratory diseases with or without wheezing were calculated. All sampled children were examined in their household at a random day by trained pediatricians using standardized procedures. The examinations included medical history, past respiratory diseases and a complete physical examination. In both surveys the examinations were distributed throughout a period of almost 12 months to assure a uniform coverage of the various areas of the city along the four seasons of the year. For each survey, the study of the social distribution of the diseases took into account tertiles of the per capita family income. For the study of the determinants of secular trends, hierarchical causal models, multivariate regression analyses and calculations analogous to the ones used to assess population attributable risks were applied. RESULTS/CONCLUSIONS: In the time span from the first to the second survey, there was a substantial increase in the prevalence of both low (from 22.2% to 38.8%) and high respiratory diseases (from 6.0% to 10.0% and from 0.8% to 2.8%, without and with wheezing, respectively). In the case of high respiratory disease and low respiratory disease without wheezing, an increase in prevalence was observed among all social strata, which did not interfere with the slightly less favourable situation of the lower income groups. In the case of low respiratory disease with wheezing, the increase was only observed among middle and low-income groups, being particularly high among the lower income group, yielding a significant inverse gradient between income and respiratory disease. Positive changes in distal (family income and maternal schooling) and in intermediate determinants related to housing characteristics would have resulted in a decline, not an increase, in the prevalence of respiratory diseases in the city. The duplication in the attendance rate to day care nurseries seen in the period could have counterbalanced the positive effect due to socioeconomic and housing variables but would not be enough to explain an increase in the disease.