Health management within couples is rarely shared equally. In cis-different-gender couples, this labor falls largely on cis-women, likely shaping how well partners know one another’s health. Same-gender couples tend to share this labor more equally, yet whether this translates into greater accuracy in proxy health reporting remains unknown. Drawing on a Gender as Relational framework, which posits that gendered health behaviors emerge from relational contexts rather than from gender alone, we examine proxy reporting congruence across four groups: men and women in cis-different-gender couples and men and women in cis-same-gender couples. Using dyadic data from the National Couples’ Health and Time Study (NCHAT), a probability-based sample of coresidential adults aged 20–60 and their partners, we restrict our sample to cisgender couples. We examine whether a partner’s proxy report matches, is near, or does not match their partner’s self-rated health, and the direction of discordance, whether partners overestimated, underestimated, or matched their partner’s self-report. Differences are estimated using multivariable multinomial logistic regression. Men in different-gender couples were more likely than all other groups to not match their partner’s self-rated health and, when discordant, to overestimate it. These patterns suggest a relational rather than individual gender effect: unequal health labor in different-gender couples may leave men less attuned to their partners’ health, whereas more equal sharing in same-gender couples may foster greater mutual awareness. These findings have implications for contexts where proxy reporting is common, including medical intake, insurance documentation, and health surveys. Systematic overestimation by men in different-gender couples could introduce a directional bias that obscures partners’ health needs. Considering the couple’s gender composition may therefore improve the interpretation of proxy-reported health data.