Income, education, occupation, age, sex, marital status, and ethnicity are all correlated with health in one context or another. This paper reflects on the difficulties encountered in deriving robust scientific conclusions from these correlations or drawing reliable policy applications. Interactions among the variables, nonlinearities, casual inference, and possible mechanisms of action are discussed. Strategies for future work are suggested, and researchers are urged to pay special attention to possible interactions among health, genes, and socio-economic variables.