Relationships between animals and their human caretakers can have profound impacts on animal welfare in farms, laboratories and zoos, while human attitudes are important predictors of caretaker behavior towards livestock. In this study, we examined the impact of keeper attitudes about working with elephants on Keeper-Elephant Relationships (KERs) and Bonds (KEBs), and found evidence for reciprocity and welfare benefits to both parties. As part of a large, multi-institutional study of zoo elephant welfare conducted at 60 zoos accredited by the American Zoo and Aquarium Association, blood samples were collected twice monthly for 1 year from 117 African (Loxodont africanus) and 96 Asian (Elephas maximus) elephants for serum cortisol analyses as a measure of well-being. Information was collected via three online questionnaires: 1) a Keeper Survey of 277 elephant keepers about their opinions of and attitudes towards working with elephants and their job satisfaction; 2) an Elephant Behavior Profile Survey where keepers rated a total of 234 elephants on the frequencies of 24 behaviors, and 3) a Keeper-Elephant Bonds Survey in which 209 individual keepers rated the strength of their bond with a specific elephant for a total of 427 keeper-elephant pairings. From the first two surveys, principle components analysis was used to create subscales of keeper attitudes, elephant behaviors and keeper job satisfaction. Component scores were then used as independent variables in epidemiological analyses of elephant mean serum cortisol and mixed model regressions of keeper job satisfaction. For African elephants, risk factors of low serum cortisol included Positive interactions with elephants (p = 0.039), Positive behaviors of elephants (friendly, affiliative) (p = 0.001), and Elephant interacts with public (p = 0.009). Age of the elephant was a small, but significant risk factor for higher cortisol (p = 0.005). For Asian elephants, the risk factors for low cortisol were attitudes indicating social inclusion in elephant groups (Keeper as herdmate, p = 0.039) and Elephant interacts with public (p = 0.006). Latitude of zoo was a predictor of higher cortisol (p = 0.041). Significant predictors of keeper Dissatisfaction with job were weaker Keeper-Elephant Bonds (p = 0.003) and African Species (p = 0.038). Species differences in KERs and KEBs are discussed in terms of differing elephant management factors in zoos. The results provide evidence of the reciprocity of KERs and the mutual benefits of KEBs to both elephant and keeper. These results are relevant for zoo animal management and staff training.