Interactions are characterized by opposite motives according to game theory. The purpose of this study was to explore how people judge the probability and advisability of conflict reactions in an unfolding dispute within a married couple using latent growth curve modeling (LGCM). Individuals participated in a study using two videotaped scenarios depicting marital conflict in which a spouse comes home after a long day at work only to criticize his or her partner for violating expectations of a good meal. One situation involved male-initiated conflict and female reactance, whereas another illustrated female-initiated conflict and male reactance. Participants were asked to predict the future reactions based on aggressive tactics (e.g., slapping the partner, insulting the partner) or prosocial and forgiving communication (e.g., apologizing, discussing the issue calmly) as well as the use of online, imagined interaction (II) rumination in which individuals replay arguments in their mind as well as thinking about what to say next during the argument. Results of the LGCM revealed support for various hypotheses in which it was predicted that the husband would be more likely to be conciliatory than the wife, and the wife would be more aggressive than her husband. II rumination was initially expected to increase and be advised before reaching a plateau. Findings are discussed in terms of game theory and II conflict-linkage theory.