Some 30 years ago, Günter Dörner proposed that exposure to hormones, metabolites and neurotransmitters during limited, sensitive periods of early development exert programming effects on disease risk in human adults. Early programming of long term health has since received broad scientific support and attention. For example, evidence increases for programming effects of infant feeding choices on later obesity risk. Meta-analyses of observational studies indicate that breast feeding reduces the odds ratio for obesity at school age by about 20%, relative to formula feeding, even after adjustment for biological and sociodemographic confounding variables. We hypothesized that breast feeding protects against later obesity by reducing the likelihood of high weight gain in infancy, and that this protection is caused at least partly by the lower protein supply with breast milk relative to standard infant formulae (the "Early Protein Hypothesis"). These hypotheses are tested in the European Childhood Obesity Project, a randomized double blind intervention trial in more than 1,000 infants in five European countries (Belgium, Germany, Italy, Poland, Spain). Formula fed infants were randomized to receive during the first year of life infant formulae and follow-on-formulae with higher or lower protein contents. Follow-up at 2 years of age shows that lower protein supply with formula normalizes early growth relative to a breast fed reference group and to the WHO growth reference. These results demonstrate that modification of infant feeding practice has an important potential for long-term health promotion and should prompt a review of the recommendations and policies for infant formula composition.