This paper investigates the predictability of divorce in a long\u2010term, prospective longitudinal study. Past research has indicated that 2 periods can be considered the most critical for the survival of marriages: (a) the first 7 years of marriage, during which half of all divorces are known to occur, and (b) the period during which the first child reaches 14 years of age, which has been suggested as a low point for marital satisfaction in the life course. In the present study, interaction variables at Time 1 (both during conflict and in an events\u2010of\u2010the\u2010day discussion following separation of the spouses for at least 8 hours) and noninteractive variables were used to predict divorcing both early and later in the marriage. A different set of variables predicted early divorcing than predicted later divorcing. Negative affect during conflict predicted early divorcing, but it did not predict later divorcing. By contrast, the lack of positive affect in events\u2010of\u2010the\u2010day and conflict discussions predicted later divorcing, but it did not predict early divorcing. Prediction was possible over the 14\u2010year period of the study with a model that included marital satisfaction, thoughts of marital dissolution, and affective interaction in both conversations. The model predicted divorce with 93% accuracy.<\/p>\n
<\/p>\n
<\/p>\n
<\/p>\n
<\/p>\n