Fasting is one of the simplest metabolic challenges that can be performed in humans. We here report for the first time a comprehensive analysis of the human "fasting metabolome" obtained from analysis of plasma and urine samples in a small cohort of healthy volunteers, using nuclear magnetic resonance (NMR), gas chromatography- and liquid chromatography-mass spectrometry (GC-MS and LC-MS). Intra- and inter-individual variation of metabolites was on measurement of four overnight fasting samples collected from each volunteer over a four week period. One additional sample per volunteer was collected following a prolonged fasting period of 36 h. Amongst a total of 377 quantified entities in plasma around 44% were shown to change significantly in concentration when volunteers extended fasting from 12 to 36 h. In addition to known markers (plasma free fatty acids, glycerol, ketone bodies) that reflect changes in the body's fuel management under fasting conditions a wide range of "new" entities such as alpha-aminobutyrate as well as other amino and keto acids were identified as fasting markers. Based on multiple correlations amongst the metabolites and selected hormones in plasma such as leptin or insulin-like-growth-factor-1 (IGF-1), a robust metabolic network with coherent regulation of a wide range of metabolites could be identified. The metabolomics approach described here demonstrates the plasticity of human metabolism and identifies new and robust markers of the fasting state.