Despite considerable effort, genetic analysis of complex disorders and traits, including those related to nutrition, has revealed only a very small part of the expected genetic variation. Missing variation may occur as conditional variation depending on the presence of defined lifestyle factors, as epigenetic variation or as low-moderate effect size variation not detected by mutation screening and genome-wide association studies. Experience with genetic analysis of patients with neural tube defects provides evidence of the existence of the latter type of variation and demonstrates that candidate gene analysis is an efficient approach to discover part of this missing variation. By reviewing the genes with rare and common variation associated with obesity or related parameters, guidelines are proposed for the proper selection of candidate genes. This selection fits into an analytic strategy to deal with the complex and massive data sets expected to arise from the application of routine whole genome sequencing.