In this study, we present a methodology for metabotyping of C. elegans using (1)H high resolution magic angle spinning (HRMAS) whole-organism nuclear magnetic resonance (NMR). We demonstrate and characterize the robustness of our metabolic phenotyping method, discriminating wild-type N2 from mutant sod-1(tm776) animals, with the latter being an otherwise silent mutation, and we identify and quantify several confounding effects to establish guidelines to ensure optimal quality of the raw data across time and space. We monitor the sample stability under experimental conditions and examine variations arising from effects that can potentially confuse the biological interpretation or prevent the automation of the protocol, including sample culture (breeding of the worms by two biologists), sample preparation (freezing), NMR acquisition (acquisition by different spectroscopists, acquisition in different facilities), and the effect of the age of the animals. When working with intact model organisms, some of these exogenous effects are shown to be significant and therefore require control through experimental design and sample randomization.