Some earlier studies based on relatively small data sets have suggested that the month of diagnosis affects survival of breast cancer patients. This phenomenon has been suggested to be attributable to daylight-related hormonal factors. Factors related to the holidays of both the medical personnel and the women themselves might also provide the explanation. In this study we assessed the effect of the month of diagnosis on the survival of 32,807 female breast cancer patients diagnosed in Finland in 1956-1985. Our results indicate that the month of diagnosis is a significant prognostic factor after adjusting for age at diagnosis, period of diagnosis, and stage at diagnosis. The adjusted relative excess risk of death was highest among those diagnosed in July and August, and lowest in March and November, the difference between the lowest and highest risk being 18%. Since colorectal cancer should not have any daylight-related hormone dependent risk determinants, a control cohort of 12,950 women with a diagnosis of colorectal cancer in the same calendar period was studied in a similar way. The survival pattern by month of diagnosis among the colorectal cancer patients was similar to that among breast cancer patients, indicating that general factors associated with the health behaviour of women and the health services (such as holidays) rather than biological factors may cause seasonal variations in survival of cancer patients.