Nutritional status of children is commonly assessed by anthropometry both in under and overnutrition. The link between anthropometry and body fat, the body compartment most affected by overnutrition, is well known, but the link with muscle mass, the body compartment most depleted in undernutrition, associated with infections, remains unknown. In this study, we examined the relationship between common anthropometric indices and body composition measured by dual-energy X-ray absorptiometry (DEXA) in a sample of 121 healthy 3-year-old Danish children. Appendicular (arms and legs) lean mass was used to estimate muscle mass. Overall, anthropometric measures were more effective to measure absolute size of fat, lean and muscle mass than their relative sizes. Proportion of the variance explained by anthropometry was 79% for lean mass, 76% for fat mass and 74% for muscle mass. For fat mass and lean mass expressed as percentage of total body mass, this proportion was 51% and 66%, respectively; and for muscle mass as percentage of lean mass it was 34%. All the best reduced multivariate models included weight, skinfold and gender except the model estimating the proportion of muscle mass in lean body mass, which included only mid-upper arm circumference and subscapular skinfold. The power of height in the weight-to-height ratio to determine fat mass proportion was 1.71 with a 95% confidence interval (0.83-2.60) including the value of 2 used in body mass index (BMI). Limitations of anthropometry to assess body composition, and especially for muscle mass as a proportion of lean mass, should be acknowledged.