Surveillance activities for ovine scrapie have expanded in the 21st century, following concerns about the potential for a hidden epidemic of bovine spongiform encephalopathy in European sheep populations. Large-scale surveys have been used to estimate the prevalence of scrapie infection. In this study we analyse data from the surveys in Great Britain between 2002 and 2004. When we estimate genotype-specific prevalences for each of the two screening tests used a difference is observed. One test underestimates the number of positive cases in genotypes classically considered to be at a low relative risk of developing clinical disease (ARR- and AHQ-containing genotypes). By comparison, the other test underestimates the number of positive cases in genotypes classically considered to be at an increased relative risk of developing clinical disease (VRQ-containing genotypes). These findings have implications for surveillance, disease control, and diagnostic test evaluation.