Polyclonal capture enzyme-linked immunosorbent assay (PC-ELISA), monoclonal capture ELISA (MC-ELISA), mouse neutralization test (MNT), and counterimmunoelectrophoresis (CIEP), were compared for their ability to detect epsilon toxin in intestinal contents and body fluids of sheep and goats. When used to evaluate intestinal contents of sheep artificially spiked with epsilon prototoxin, PC-ELISA detected 0.075 mouse lethal dose (MLD)50/ml, whereas the MNT, MC-ELISA, and CIEP detected 6, 25, and 50 MLD50/ml, respectively. Amounts of epsilon toxin detected by PC-ELISA, MC-ELISA, MNT, and CIEP in sheep pericardial fluid artificially spiked with epsilon prototoxin were 0.075, 0.75, 6, and 200 MLD50/ml, respectively. For assaying epsilon toxin in aqueous humor, PC-ELISA and MC-ELISA detected 0.075 MLD50/ml, whereas CIEP detected 200 MLD50/ml (MNT was not evaluated). When 51 samples of intestinal contents of sheep and goats (32 positive and 19 negative to MNT) were analyzed by the other 3 techniques, the relative sensitivity of PC-ELISA, MC-ELISA, and CIEP was 93.75, 84.37, and 37.50%, respectively. The specificity of PC-ELISA, MC-ELISA, and CIEP was 31.57, 57.89, and 84.21%, respectively. The absolute sensitivity of PC-ELISA, MC-ELISA, CIEP, and MNT was 90.90, 69.69, 15.15, and 54.54%. The absolute specificity of the 4 techniques was 100%. These results show that there is a marked inconsistency among techniques routinely used to detect Clostridium perfringens epsilon toxin. Until more consistent results are achieved, the diagnosis of enterotoxemia should not only be based solely on epsilon toxin detection, but also on clinical and pathological data.