Marital disruption is associated with increased risk for a range of poor health outcomes, including disturbed sleep. This report examines trajectories of actigraphy-assessed sleep efficiency following marital separation as well as the extent to which daily social behaviors and individual differences in attachment explain variability in these trajectories over time. One hundred twenty-two recently-separated adults (N = 122) were followed longitudinally for three assessment periods over 5 months. To objectively assess daily social behaviors and sleep efficiency, participants wore the Electronically Activated Recorder (EAR) during the day (for one weekend at each assessment period) and an actiwatch at night (for 7 days at each assessment period). Greater time spent with an ex-partner, as assessed by the EAR, was associated with decreased sleep efficiency between participants (p = .003). Higher attachment anxiety was also associated with decreased sleep efficiency (p = .03), as was the EAR-observed measure of “television on.” The latter effect operated both between (p = .004) and within participants (p = .005). Finally, study timepoint moderated the association between EAR-observed television on and sleep efficiency (p = .007). The current findings deepen our understanding of sleep disturbances following marital separation and point to contact with an ex-partner and time spent with the television on as behavioral markers of risk.