To examine the relationship between stress and upper respiratory tract infection, 235 adults aged 14-57 years, from 94 families affiliated with three suburban family physicians in Adelaide, South Australia, participated in a six-month prospective study. High and low stress groups were identified by median splits of data collected from the Life Events Inventory, the Daily Hassles Scale, and the General Health Questionnaire, which were administered both before and during the six months of respiratory diary data collection. Using intra-study stress data, the high stress group experienced significantly more episodes (mean of 2.71 vs. 1.56, p less than 0.0005) and symptom days (mean of 29.43 vs. 15.42, p = 0.005) of respiratory illness. The two groups were almost identical with respect to age, sex, occupational status, smoking, passive smoking, exposure to air pollution, family size, and proneness to acute respiratory infection in childhood. In a multivariate model with total respiratory episodes as the dependent variable, 21% of the variance was explained, and two stress variables accounted for 9% of the explained variance. Significant, but less strong relationships were also identified between intra-study stress variables and clinically "definite" episodes and symptom days in both clinically definite and total respiratory episodes. Pre-study measures of stress emphasized chronic stresses and were less strongly related to measures of respiratory illness than those collected during the study. However, significantly more episodes (mean of 2.50 vs. 1.75, p less than 0.02) and symptom days (mean of 28.00 vs. 17.06, p less than 0.03) were experienced in the high stress group. In the multivariate analyses, pre-study stress remained significantly associated with total respiratory episodes and symptom days in total and "definite" respiratory episodes. In all of the multivariate analyses performed, sex (female) and age also appeared as important correlates of respiratory illness.