Two dietary reporting methods were used to examine associations between macronutrient intake and subsequent cancer incidence in a cohort of Swedish women born between 1908 and 1930. 1361 subjects gave 23-h dietary recalls at their baseline examinations in 1968-1969, and 412 of them also provided detailed dietary histories. The cohort was followed up 19 years later by means of linkages with the National Cancer and Death Registries. Both dietary methods indicated that subjects who were ranked in the highest tertile of energy intake, relative to the lowest, were at significantly greater risk of developing cancer (all-site). Relative risks across energy intake tertiles were 1, 1.15 and 2.04, respectively, using the dietary history method and 1, 1.02 and 1.55 using the 24-h recall data. Examination of specific macronutrient energy sources indicated that dietary fat and carbohydrate are likely to have made the largest contribution to this association. However, after adjustment for total energy, none of the individual macronutrients was significantly associated with all-site cancer, by either dietary reporting method. When expressed as a percentage of total energy, low protein density of the diet was associated with increased cancer risk, by both dietary methods. However, this appeared to be a function of high energy intake rather than low protein intake. Simultaneous statistical adjustment for seven potential confounders of the association between energy intake and cancer was performed using both the 24-h recall and the dietary history data.(ABSTRACT TRUNCATED AT 250 WORDS)