The overall objective of this research was to create a welfare index for captive elephants for uniform assessment of welfare and husbandry standards that would enable captive elephant facilities to identify and improve areas of concern in elephant management and welfare. Secondary objectives were to assess the welfare of captive Asian elephants (Elephas maximus) through surveys and non-invasive measures and determine public attitudes to elephants in zoos. Expert opinion and interviews initially identified 15 important welfare attributes with three to six levels of each. An online survey was then conducted among five stakeholders groups (n = 242) to identify the relative importance of each attribute, which created the following ranking of attributes in declining order of importance: substrate > group size > health care > enrichment > restraint > enclosure type > exercise provision > enclosure size > interaction with keeper > enclosure environment > keeper knowledge/experience > diet > handling methods > display duration > enclosure security. Out of a maximum of 10, the importance scores for the attributes ranged from 8.14 to 5.23. The importance scores for attributes and the utility values of each level were combined to construct a captive elephant welfare index, which can be used to create an overall index score for any elephant enclosure. This can be used to ascertain the welfare status of animals in that enclosure and identify areas of management inadequacies for improvement. To strengthen the usefulness of the welfare index and comprehensively estimate the welfare of captive elephants, further studies on behaviour and cortisol were conducted in three sanctuaries and three zoological parks in southern India to correlate the index score with the observed findings. The welfare index scores tended to correlate negatively with cortisol concentrations in the urine (P = 0.06), and with time spent head weaving (p = 0.001) and walking typical of musth (p = 0.02) behaviours. Idling time, which included sleep, correlated positively with the index score (p = 0.004). These results together indicate that the welfare index was correlating with commonly used measures of welfare. Cortisol concentrations in the urine of elephants in sanctuaries (277 ± 42.7 ng/mg Cr, p = 0.002) was less than that of elephants in zoological parks (402 ± 73.9 ng/mg Cr, p = 0.002), indicating reduced stress in sanctuaries. Furthermore, elephants in sanctuaries performed two types of behaviour, manipulating enrichment (p = 0.02) and walking (p = 0.01), more than elephants in zoological parks. Stereotypic behaviours, including body swaying (p = 0.002), and repetitive trunk movements (p = 0.01), were found to be significantly higher in zoos than sanctuaries. Cortisol values showed a negative correlation with the walking behaviour (p < 0.001) and a positive correlation with the behaviours idling (p = 0.02) and dusting (p = 0.03). Thus it was concluded that elephants in sanctuaries had better welfare conditions than those in zoological parks. Further to assess the validly and usefulness of the welfare index, an email questionnaire sent to different sanctuaries and zoological parks, containing elephant enclosures worldwide, producing 37 completed responses.. The survey gathered information about the husbandry and management conditions of these institutions and assigned index scores for each of the elephant enclosures, based on the levels of each attribute that was self-selected by the respondents. The mean welfare index score achieve by the all the zoos and sanctuaries was 77.4 %, with the maximum score being 87 % and the minimum 55 %. Zoos had most trouble meeting adequate standards of enclosure substrate, social structure, exercise provision and interaction with keepers, whereas they generally performed well in health care, size and type of enclosure and diet. Public attitudes towards the welfare of elephants in captive enclosures were determined through a telephone survey administered in southern India (n = 101) and Australia (n= 101). Australian’s concerns about elephants were more than those of Indians (p = 0.02). A total of 42.6 % of Australian respondents were willing to pay more to visit a zoo having elephants as against 7.9 % of the Indian respondents, if that extra money was for improving the welfare conditions of the elephants. Overall the respondents from India considered elephants to be more important in terms of cultural, religious and historical significance, whereas Australians considered them to be more important in scientific value. This research has developed a welfare index for elephants, validated it using behavioural and physiological data obtained in the field, and evaluated performance of zoos worldwide, as well as comparing attitudes to elephants in zoos in two contrasting countries.