Studies examining the effects of positive psychological interventions (PPIs) in populations with diverse sociodemographic profiles are needed. We tested the impact of a 6-week, telephone-based PPI (vs. active control program) among United States Military Veterans with chronic knee pain. The parent study was powered to detect racial differences in pain symptoms over time. For this analysis, we used mixed model linear regression to test racial and gender differences in effects of the PPI (vs. control) over time on positive affect, negative affect, and affect balance (positive minus negative affect). Randomized participants (N=517, mean age=63.7, 52% African American, 48% White, 27% female) completed the 10-item International Positive and Negative Affect Schedule-Short Form at baseline and 1, 3, and 6 months post-intervention. We found significant 3-way interactions between race, study arm, and time for positive affect and affect balance, but not for negative affect. Significant interactions were driven by an increase in positive affect from baseline to 1 month among White control participants, with a corresponding decrease in positive affect among White PPI participants. Gender interactions were not significant for any measures. Potential implications of these findings for measuring affect in PPI studies among unique subpopulations will be discussed.