Current research focusing on long-term marriages have assessed marriages at one time period only. To sort out age and stage differences from cohort effects, a longitudinal case study analysis was undertaken with the 17 surviving couples of the Berkeley Older Generation Study, a group that has been interviewed in young adulthood (1929-30), in middle age (1945-47), in young-old age (1968-69), and in old-old age (1982-83). In order to trace the course of marital satisfaction and to identify factors affecting this course, socioeconomic status, health, availability of children, distribution of power, decision-making and conflict resolution styles, and affective components from the early years of marriage to the final years were examined. Seven marriages were classified as intrinsic, six as conflicted, three as passive-congenial, and one as hostile-accommodating. All of the marriages were evolving and developing, changing with life stage, circumstances, and time. For men, the number of things done together, the importance of sex, and thoughts of divorce were correlated with marital satisfaction. For women, factors important to marital satisfaction varied considerably from one time period to another (child-related variables, work, education, personality, life satisfaction). By old-old age marital satisfaction for men and women was highly related to the number of disagreements. (NRB)