Relationships between dietary nutrients and plasma prolactin concentration were studied in 249 women with a history of nonskin cancers among first-degree female relatives. For each quintile of nutrient density, the odds ratio (OR), relative to the lowest quintile, of having an elevated (above the median) prolactin concentration was estimated by logistic regression, taking into account parity, menopausal status, and current tobacco-smoking habits. For nutrient densities estimated from 24-h recall data there was a significant positive association between plasma prolactin concentration and increasing saturated fatty acid intake; the OR of elevated prolactin in the top quintile was 3.1 [95% confidence interval (CI) 1.2-8.1] and there was a negative association with vitamin C [OR in the top quintile 0.28, (95% CI 0.10-0.78)]. For usual nutrient densities (estimated by quantitative food frequency questionnaire) there was a statistically significant trend (P = 0.04) toward lower prolactin concentrations with increasing sodium density, and a marginally significant positive trend (P = 0.07) with increasing dietary density of refined sugars.