Objective: To describe infant feeding practices in rural Senegal in relation to age and nutritional status. The main hypothesis to be tested was whether mothers modulate feeding in response to growth and nutritional status of their infants. Design: A cross-sectional survey using qualitative 24-h recalls and lifetime recalls to assess feeding practices, and using weight and recumbent length measurements to assess nutritional status. Setting: Three health clinics in the Fatick region, a rural area of Senegal, West Africa, covering a population of 26 600. Subjects: All 2-10-month-old infants attending four immunization sessions in 1991 (n = 1174; 80% of convoked infants). Main outcome measures: Prevalence of feeding with additional food (gruel, family diet and food of animal origin), weight-for-length and length-for-age. Results: All infants were breastfed. A supplement had been given the day before the survey to 10% of infants aged 2-3.9 months, 30% of infants aged 4-5.9 months and 45% of those aged 6-6.9 months. The main food items were watery millet gruel and family diet (millet or rice). Gruel was given in response to perceived breast-milk insufficiency. Animal products were seldom eaten at any age. Length-for-age and weight-for-length were significantly lower among infants supplemented with millet gruel, when adjusted for age; while no such relationship was found with family diet. Conclusion: Mothers preferentially fed gruel to small, thin infants.