Abstract: AIMS: To investigate the utility of faecal oestrone sulphate (OS) concentrations for detecting pregnancy in mares during behavioural studies of feral horses, in which the collection and preservation of samples is not immediate. METHODS: Oestrone sulphate concentrations were measured in fresh dung samples collected from 153 free-roaming Kaimanawa mares throughout the year. In addition, multiple samples were taken from the same pile to investigate the reliability of diagnosis from a single sample, as well as the influence of time until preservation on OS concentrations. Samples were also taken before and after a 10mm simulated rainfall event to test for dilution of OS concentrations by rain. Oestrone sulphate concentrations in all samples were measured using an enzyme immunoassay. RESULTS: From approximately 150 to 250 days of gestation, OS concentrations were consistently >80 ng/g in mares which subsequently foaled. Mares which did not foal and had low faecal OS concentrations in multiple samples throughout the year had faecal OS concentrations of 31+/-13 ng/g (mean+/-s.d.) with an upper 95% confidence limit of 57 ng/g. Mares sampled from 1 week before to 1 month after behavioural oestrus, and that did not foal in the previous and subsequent seasons, had OS concentrations of 37+/-32 ng/g (mean+/-s.d.) with an upper 95% confidence limit of 100 ng/g. The standard error of oestrone sulphate concentrations in multiple samples from the same dung pile ranged from 1 to 37% of the mean. This large within-pile variation, however, did not result in incorrect diagnoses from single samples unless mares were within 18 days of parturition. Keeping samples at ambient temperatures for up to 16 hours did not affect OS concentrations. Simulated rainfall caused a 17% mean reduction in OS concentrations, but did not change pregnancy diagnoses. CONCLUSIONS: Faecal OS concentrations >100 ng/g were indicative of pregnancy in Kaimanawa mares. For mares more than 150 days post-mating, OS concentrations <57 ng/g were indicative of non-pregnancy, while concentrations between 57 and 100 ng/g provided an inconclusive diagnosis. A single sample from each dung pile collected within 16 hours of defecation was sufficient to accurately diagnose pregnancy in mares 150-250 days post conception. CLINICAL RELEVANCE: Measurement of OS concentrations in dung samples was a reliable and robust indicator of pregnancy status in feral mares 150-250 days post mating. This corresponds approximately to the period from May to August, given the seasonal breeding pattern in this population. This method of determining pregnancy status is suitable for field use in behavioural and demographic studies of wild horse populations.