OBJECTIVE: To assess the relation between self reported intensity and frequency of stress and first time incidence of primary breast cancer. DESIGN: Prospective cohort study with 18 years of follow-up. SETTING: Copenhagen City heart study, Denmark. PARTICIPANTS: The 6689 women participating in the Copenhagen City heart study were asked about their perceived level of stress at baseline in 1981-3. These women were followed until 1999 in the Danish nationwide cancer registry, with < 0.1% loss to follow-up. MAIN OUTCOME MEASURE: First time incidence of primary breast cancer. RESULTS: During follow-up 251 women were diagnosed with breast cancer. After adjustment for confounders, women with high levels of stress had a hazard ratio of 0.60 (95% confidence interval 0.37 to 0.97) for breast cancer compared with women with low levels of stress. Furthermore, for each increase in stress level on a six point stress scale an 8% lower risk of primary breast cancer was found (hazard ratio 0.92, 0.85 to 0.99). This association seemed to be stable over time and was particularly pronounced in women receiving hormone therapy. CONCLUSION: High endogenous concentrations of oestrogen are a known risk factor for breast cancer, and impairment of oestrogen synthesis induced by chronic stress may explain a lower incidence of breast cancer in women with high stress. Impairment of normal body function should not, however, be considered a healthy response, and the cumulative health consequences of stress may be disadvantageous.