We compared the use of recto-anal mucosal swab (RAMS) culture and faecal culture for the detection of E. coli O157 in a mob of Merino sheep. Fifty Merino wethers and maiden ewes housed in indoor pens were sampled on five occasions. We detected E coli O157 in 32% (16/50) of sheep, with weekly prevalence ranging from 4% (2/50) to 16% (8/50). Overall, 12·5% (2/16) were detected by RAMS culture only, and 37·5% (6/16) were detected by faecal culture only. The level of agreement between the two sampling methods was moderate [kappa statistic = 0·583, 95% confidence interval (CI) 0·460–0·707]. The relative sensitivities of RAMS and faecal culture were 67% (95% CI 41–86) and 57% (95% CI 34–77), respectively. We identified four super-shedding sheep using direct faecal culture. Although the majority of culture-positive sheep were detected at one sampling point only, 3/4 super-shedding sheep were culture-positive at two sampling points, and 1/4 was culture-positive at four sampling points. Persistent culture positivity may indicate sheep that could be considered ‘super-shedders' at some point. The use of immunomagnetic separation further improved the rate of detection of E. coli O157, which was isolated from 1/34 animals that were previously negative by enrichment culture alone. A significant difference between sampling weeks was detected for both faecal (P = 0·021) and RAMS (P = 0·006), with the prevalence at the mid-point of sampling (week 4) significantly (P < 0·05) higher than at the beginning or end of the study. Study conditions (penned sheep) might have been responsible for the high prevalence and the epidemic pattern of infection observed, and could serve as a future model for studies of E. coli O157 transmission, shedding and super-shedding in sheep.