The paper develops an overlapping generations model with probabilistic aging of households. We define age as a set of personal attributes such as earnings potential, health and tastes that are characteristic of a person's position in the life-cycle. In assuming a limited number of different states of age, we separate the concepts of age and time since birth. Agents may retain their age characteristics for several periods before they move with a given probability to another state of age. Different generations that share the same age characteristics are aggregated analytically to a low number of age groups. The probabilistic aging model thus allows for a very parsimonious yet rather close approximation of demographic structure and life-cycle differences in earnings, wealth and consumption. Existing classes of overlapping generations models follow as special cases.