Epidemiological studies of aging are usually confronted with the presence of numerous pathologies or environmental factors which make it difficult to identify the effects of aging individually. One way of reducing the variability among individuals is to use well defined criteria to select the study population. This is the choice that was made for the New Mexico and Toulouse Aging Process Studies, which were particularly turned towards successful aging. In this study we have sought to explain the diversity of states of health of the subjects of these two studies by means of an aging classification exploring the medical history, balance and walking, and the cognitive functions. This reveals that the poorer health of certain subjects (about 10% and 30% of the populations of Toulouse and Albuquerque respectively) is slightly associated with changes in eating behaviour relative to subjects who age successfully. We have, however observed a decrease in vitamin E intake in Toulouse associated with a reduction of lipids in the food. But the poorer state of health is predominantly explained by advancing age and the occurrence of pathological states. This study confirms that aging was generally successful in the two populations studied.