Stool culture for Clostridium difficile, while necessary for strain typing and antimicrobial surveillance, cannot determine toxin production. We prospectively tested in triplicate 91 C difficile cultured isolates for toxin production by 2 enzyme immunoassays (EIAs) (Meridian Premier Toxins A&B, Meridian Bioscience, Cincinnati, OH; and TechLab Tox A/B II, TechLab, Blacksburg, VA) and cytotoxin neutralization bioassay (CTN). By CTN, 88% (80/91) were toxigenic. Reproducibility was 93% (85/91) for CTN, 80% (73/91) for Meridian EIA, and 79% (72/91) for TechLab EIA. Compared with CTN, sensitivities were 87.1% and 89.2% for the Meridian and TechLab EIAs, respectively. In an additional 115 stool specimens, CTN detected toxin more frequently from cultured isolates (96/115) than stool (84/115). For C difficile toxin detection from isolates, EIA was less reproducible than CTN. EIA methods can be falsely negative in 10% to 12% of isolates, and these should be tested by CTN or polymerase chain reaction. When positive, EIA is fast and reliable for detecting C difficile toxin from culture.