Blood donation with compensation is considered as a social stigma. However, more people in the reference group donate blood often leads to less moral concern and more followers. Therefore, the behavior is likely to be influenced through one’s interactions with neighbors, friends and relatives. Meanwhile, relative income may affect the motives for blood donation through increasing mistrust and stress. The motives might be stronger for households of lower social rankings. Utilizing three-wave census-type panel data in 18 villages in rural western China, two identification strategies, instrumental variable and network-based identification, are implemented to estimate the effect of social interactions. Both community-specific and household-specific relative income measures are employed to test whether blood donation is more sensitive towards the less well-off in a society. We find strong evidence in support of the effects of social interactions, no matter whether instrumental variables or network centrality measures are adopted. Household-specific measures of relative income show more salient effects on blood donation than community-specific inequality.