Bovine spongiform encephalopathy (BSE) may have been transmitted to British sheep via contaminated feed in the 1980s. Strain-typing techniques based on immunohistochemical (IHC) detection of abnormal protein (PrP(d)) and the molecular analysis of proteinase-resistant protein (PrP(res)) by Western blotting (WB) can discriminate between natural or experimental scrapie and experimental BSE in sheep. Between 1 January 1998 and 31 October 2001, 1247 sheep, clinically suspected of scrapie, were found to be positive by statutory tests in Great Britain. Archived brain tissue from these cases was retested by using these discriminatory methods. Twelve brain samples showed PrP(res) WB patterns that were unlike those found in natural or experimental scrapie. Prospective screening of fresh tissue from a further 1121 scrapie cases was also carried out between 1 November 2001 and 31 May 2004. Two samples gave WB results with similarities to the results found for experimental BSE in sheep. When all 14 unusual cases were tested by IHC, no match to experimental BSE in sheep was found. There were uncertainties within the retrospective study, where some equivocal results were obtained due to poor tissue quality or the unavailability of the optimum brain region. However, for the samples where tissue condition was optimum, our results provide no evidence for the presence of BSE in sheep. Epidemiological interpretation of the 450 flocks sampled indicates that the maximum proportion of sheep transmissible spongiform encephalopathy cases that could be BSE is 0.66%. This estimate is lower than calculated previously (5%), when the analysis was based on the results of strain typing in mice.