In 1998, 30 juvenile elephants were captured in Botswana and transported to a holding facility in South Africa to be trained and sold to zoos and safari parks. The welfare of the 'Tuli Elephants' as they became known, became the source of acrimonious dispute between a number of conservation and animal groups. The case highlighted concerns over the welfare of elephants during capture, transport and confinement. Questions asked were: can an objective assessment of the effect on juvenile elephants on the removal from matriarchal group be made. This study was aimed to take an objective approach to assessing the welfare of captive juvenile elephants using behavioural and physiological methods of investigation. A behavioural study to identify indicators of stress was conducted in five groups of elephants subject to various husbandry systems. Thirteen behavioural indicators of stress were identified. A group of two elephants held in an enclosure 70 m2 that was devoid of mud and sand baths showed the highest number of behavioural indicators of stress. Elephants in larger enclosures with mud and sand baths showed fewer indicators of stress. The group able to range freely during the day showed the least number of stress¬related indicators. Conventionally the physiological assessment of adrenal responses to stress relies upon blood sample collection and the measurement of glucocorticoids but this is impossible without immobilisation or restraint that influences results. This study validated a recently established enzyme immunoassay (EtA) measuring faecal glucocorticoid metabolites in elephants. A preliminary investigation into the biological relevance of this non-invasive method was made for use in assessing welfare in elephants. Four juvenile elephants were injected i.m. with synthetic adrenocorticotrophic hormone (ACTH) (Synacthen, Novartis; 2.15 mg). Blood and faecal samples were collected over 4 hand 7 days respectively. Concentrations of serum cortisol and faecal cortisol metabolites were determined using immunoassay. Variability of basal and peak values in blood and faeces were observed among the elephants. After ACTH injection, serum cortisol concentrations increased by 400-700%. When compared to cortisol and corticosterone EIAs, 11-oxoaetiocholanolone EIA proved best suited to measure cortisol metabolites. Concentrations of faecal 11,17¬dioxoandrostanes increased by 570-1070% reaching peak levels after 20-25.5h. Samples left outside could be collected up to 8hrs after defecation without a significant effect on metabolite concentrations. A correlation between enclosure size, presence of stress-related behaviour and faecal 11,17 -DOAs was observed. Elephants kept in small enclosures exhibited more stress-related behaviour and had higher levels of glucocorticoid metabolites than those ranging in a larger area. The results of the study suggest that non-invasive faecal monitoring of glucocorticoid metabolites is useful in investigating adrenal activity in African elephants.