Objective: To study energy and macronutrient intake in relation to dental caries incidence among urban black South African children at the ages of 1 and 5 years. Design and subjects: Food frequency questionnaires and WHO caries diagnostic criteria were used. The study sample of 1639 urban black children with nutrition and dental information from the 1991 and/or 1995 Birth-to-Ten (BTT) Study interceptions comprised true longitudinal (n = 259) and cross-sectional (n = 1216 for 1991 and n = 164 for 1995) subjects. For the longitudinal group true dmfs (decayed, missing, ®lled surfaces) incidence and average intakes of energy and macronutrients between 1991 and 1995 were calculated. Setting: South Africa. Results: Median macronutrient intakes were not signi®cantly different between the cross-sectional and longitudinal samples. Dental caries prevalences at age 1 were 1.2% and 1.5% for the cross-sectional and longitudinal samples, respectively. By age 5 these values were 60.4% and 62.2%, respectively. The median dmfs score at age 1 was 0. At 5 years this was 2 for all children and 5 for those with caries. Statistical analysis for trend at age 5 showed a signi®cant increase in caries prevalence with increasing energy, carbohydrate and added sugar in the cross-sectional sample only. The only signi®cant changes in dmfs scores were seen for energy and added sugar for the whole sample. The r values between log(dmfs incidence 0.3) and the average nutrient log variables were low, and a general linear models analysis showed no statistically signi®cant effects on log(dmfs incidence 0.3) of any of the log nutrient variables. Conclusions: For the relationship between macronutrient intake and caries (prevalence and incidence), a cross-sectional and true longitudinal study gave similar results.