This article provides definitions, concepts, and a general description of event history analysis relative to its potential application to marital and family interaction data. It discusses the differences between continuous-time and discrete-time analysis and the differences between parametric and proportional hazards models. Finally, the article addresses data collection and analytic issues relevant to the family interaction investigator.Suppose that you are interested in examining mother and son conflict episodes because previous research suggests that the duration and outcome of these short microbursts of interaction predict the son's subsequent development of antisocial behavior patterns (Patterson, 1982). The hypothesis is simple: The more frequently conflict occurs and the longer, on average, each conflict lasts, the worse the prognosis for the son's future. The first part of the hypothesis is a density issue, and the second is a rate-of-change issue. Suppose further that you want to determine what measurable factors influence the likelihood of conflict as well as how rapidly the dyad moves from the conflict state to the nonconflict state. Alternatively, you might ask what factors influence how rapidly the dyad moves to a conflict state if they are currently in a nonconflict state.Although several common data-analytic strategies are now available to address the initial part of the hypothesis, which is the event likelihood problem (e.g., logistic regression), analyzing the second part, or the rate of change, is more difficult. Until recently, such questions were typically either not asked, or duration was averaged by variable level across multiple variables, and then a comparison procedure determined the presence of statistically significant differences. Aside from the crudeness of such a procedure, an additional problem is that it portrays the interaction between time and behavior as static,