BACKGROUND: Maternal undernutrition leads to an increased risk of metabolic disorders in offspring including obesity and insulin resistance, thought to be due to a programmed thrifty phenotype which is inappropriate for a subsequent richer nutritional environment. In a rat model, both male and female offspring of undernourished mothers are programmed to become obese, however postnatal leptin treatment gives discordant results between males and females. Leptin treatment is able to rescue the adverse programming effects in the female offspring of undernourished mothers, but not in their male offspring. Additionally, in these rats, postnatal leptin treatment of offspring from normally-nourished mothers programmes their male offspring to develop obesity in later life, while there is no comparable effect in their female offspring. RESULTS: We show by microarray analysis of the female liver transcriptome that both maternal undernutrition and postnatal leptin treatment independently induce a similar thrifty transcriptional programme affecting carbohydrate metabolism, amino acid metabolism and oxidative stress genes. Paradoxically, however, the combination of both stimuli restores a more normal transcriptional environment. This demonstrates that "leptin reversal" is a global phenomenon affecting all genes involved in fetal programming by maternal undernourishment and leptin treatment. The thrifty transcriptional programme was associated with pro-inflammatory markers and downregulation of adaptive immune mediators, particularly MHC class I genes, suggesting a deficit in antigen presentation in these offspring. CONCLUSIONS: We propose a revised model of developmental programming reconciling the male and female observations, in which there are two competing programmes which collectively drive liver transcription. The first element is a thrifty metabolic phenotype induced by early life growth restriction independently of leptin levels. The second is a homeostatic set point calibrated in response to postnatal leptin surge, which is able to over-ride the metabolic programme. This "calibration model" for the postnatal leptin surge, if applicable in humans, may have implications for understanding responses to catch-up growth in infants. Additionally, the identification of an antigen presentation deficit associated with metabolic thriftiness may relate to a previously observed correlation between birth season (a proxy for gestational undernutrition) and infectious disease mortality in rural African communities.