Background: Moderator analyses may be helpful for evaluating intervention effects. The objective of this study was to evaluate whether the effect of a caregiver skills training intervention-Helping Invested Families Improve Veterans' Experience Study (HI-FIVES)-on care recipient outcome and caregiver outcome is moderated by the veteran's risk for hospitalization or level of functional impairment. Methods: Secondary data analysis of HI-FIVES. Outcomes included veteran days in the community (cumulative days in the community at 12 months alive and not in the emergency department, hospital, or post-acute facility) and caregiver burden (Zarit Burden Interview at baseline, 3, 6, and 12 months). Moderators, risk for hospitalization and functional impairment, was assessed using their Care Assessment Need (CAN) score and the Older Americans Resources and Services (OARS) Questionnaire, respectively. Negative binomial and linear mixed models were used to assess intervention effects on community days and caregiver burden, respectively. Results: Of 241 dyads, veteran (caregiver) mean age was 73 (61) years, 53% (54%) were Black, and 4% (89%) were female. Medium hospitalization-risk (91-98 CAN score) participants who received HI-FIVES had 3.75 times more community days vs. medium risk participants in the control group. By contrast, the high hospitalization-risk (99+ CAN score) participants who received HI-FIVES had 4.39 times fewer community days vs. high-risk participants in the control group. The CAN score did not have a significant moderating effect on caregiver burden. OARS did not have a significant moderating effect on community days or caregiver burden. Conclusions: Targeting the HI-FIVES based on hospitalization risk may be an effective strategy for keeping veterans in the community longer.