Objectives: We theorize and test the health of older adults as a result of their activity engagement, as well as a product of their spouse's engagement. Method: We draw on 15 waves of couple-level data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. Using responses of time engaged in nine different activities, we estimate Latent Class Models to describe activity profiles of partnered older adults. Given potential health selections into activity engagement, we lag older adults' activity engagement by one wave to examine its association with subsequent health. We then investigate associations between the lag of the spouse's activities with respondents' health, controlling for their own activity engagement at the previous wave. Result: We find four activity profiles for men, and three for women. Respondents who were predominantly engaged in community activities generally report better subsequent health. Beyond their own activity engagement, for both older men and women, having a partner who was also community engaged associate with better subsequent health, though for older women, there were little differences between having a husband who was community engaged or inactive. Discussion: Our findings highlight the value of considering activities of partnered older adults at the couple level.