A longitudinal framework was used to examine the competing hypotheses of (a) whether family functioning predicts changes in posttraumatic stress disorder (PTSD) symptoms or (b) whether PTSD symptoms predict changes in family functioning. Veterans (N = 311) admitted to a treatment program completed a series of questionnaires at 3 time points: at intake, from intake to completion of a treatment program, and at the 6-month follow-up. Alcohol use and general mental health symptoms were also measured at intake. A cross-lagged panel model using structural equation modeling analyses indicated that family functioning was a moderate predictor of PTSD symptoms at posttreatment and at the 6-month follow-up. PTSD was not a significant predictor of family functioning across time and alcohol use, and general mental health symptoms did not affect the overall findings. Further analyses of PTSD symptom clusters indicated that the avoidance symptom cluster was most strongly related to family functioning. Targeting family relationships for treatment may be important in the future for veterans with PTSD.