Background: Research suggests bidirectional associations between symptoms of posttraumatic stress disorder (PTSD) and sleep disturbance, both of which have been associated with alcohol problems. However, few studies have examined the interplay of these conditions in predicting alcohol problems over time. This study tested 2 competing models: (i) sleep disturbance as a mediator of the association between intrusive thoughts about trauma and alcohol problems and (ii) intrusion symptoms as the mediator of the sleep/alcohol problem association.Methods: Veterans (N = 325, 93% male, 81% White) completed assessments at baseline, 6 months, and 12 months as part of a larger observational study. Zero-inflated-negative binomial models were used to examine indirect effects of baseline predictors on (i) yes/no likelihood and (ii) number of 12month alcohol problems through 6-month mediators. Models controlled for past-year cannabis use and drinks consumed per week at baseline.Results: The only significant predictor of alcohol problem likelihood was baseline drinking quantity. Baseline PTSD intrusions had a direct effect on number of alcohol problems at 12 months, with no indirect (mediated) effect through 6-month sleep disturbance. In the competing model, baseline sleep disturbance had a marginally significant direct effect on 12-month alcohol problems, with a significant indirect effect through 6-month PTSD intrusions.Conclusions: PTSD intrusions are associated with more alcohol problems and help explain the long-term association between sleep and alcohol problems among veterans. Because sleep disturbances are associated with more intrusive thoughts about trauma, we recommend that treatments targeting sleep in the context of PTSD and alcohol use include a cognitive component.