The agreements between various culturing-method combinations and specimens to detect sheep naturally infected by Salmonella enterica IIIb 61:k:1,5,(7) were tested. Rectal swabs, faecal samples and ileo-caecal lymph nodes were collected from each individual. Rectal swabs called "group I" (n = 54) were cultured directly on selective media (selenite cysteine, SC). Rectal swabs called "group II" (n = 47) were pre-incubated in buffered peptone water. The four other combinations of culturing-method and specimen (one lymph node and three faecal) in each of the two groups were cultured directly on SC. The results from all the combinations of method and specimens were compared to the result from the rectal-swab test by kappa and McNemar's chi-square. Kappas for the agreement between rectal swabs and faecal sample tests were only 0.4. The lymph-node test agreed even-more poorly, with mean kappa value -0.02. The McNemar's test revealed that the discordancies between the rectal-swab test and all of the other tests went in both directions but sometimes did show bias (directionality) in the discordancies.