The aging phenotype in humans is very heterogeneous and can be described as a complex mosaic resulting from the interaction of a variety of environmental, stochastic and genetic-epigenetic variables. Therefore, each old person must be considered as a singleton, and consequently the definition of 'aging phenotype' is very difficult. We discuss the phenotype of centenarians, the best example of successful aging, as well as other models exploited to study human aging and longevity, such as families enriched in long-living subjects, twins and cohorts of unrelated subjects. A critical review of literature available until March 2008. No single model can be considered the gold standard for the study of aging and longevity, instead the combination of results obtained from different models must be considered in order to better understand these complex phenomena. We propose that a systems biology concept such as that of 'bow-tie' architecture, useful for managing information flow, could help in this demanding task.