The association between self-reported intake of various dietary factors at diagnosis and survival from breast cancer was studied in a population-based cohort of breast cancer patients in Adelaide, South Australia. These patients had been recruited between 1982 and 1984 into a case-control study of diet and incident breast cancer. Of the 451 patients recruited originally, 412 were followed for a median interval of 5.5 years. There were decreases in the risk of death from breast cancer ranging from 25 to 40% at all levels of energy and protein intake above the baseline, whereas for fat intake there was a 40% increase in risk at the uppermost quintile level. There was also some reduction in risk at the upper levels of intake of beta-carotene and vitamin C. However, there were no dose-dependent variations in risk of death by level of intake for any of the dietary factors studied, and most of the variation in risk that was observed was relatively insubstantial.