The aim of this study was to compare different methods for C. difficile toxins detection. Fifty three stool samples taken from patients with antibiotic-associated diarrhoea were studied. TCD toxin A EIA (Becton Dickinson, USA), Tox A/B ELISA test (TechLab, USA), cytotoxicity and neutralization assay on McCoy cells and PCR for detection of both toxin A and B genes were performed in vivo (in stool samples) and in vitro (in isolated strains). Reference toxigenic and nontoxigenic and two Japanese toxin A-negative and toxin B-positive C. difficile strains were used as a controls. TCD toxin A EIA detected in vivo only 19 positive samples. Tox A/B test detected 52 positive samples out of 53 studied. All 53 stool samples were C. difficile culture positive (53 strains were cultured). Toxin B was detected in 52 strain-supernatants and in all controls (except the nontoxigenic one). Both toxin A and B genes were detected by PCR in all 53 isolated strains, Japanese and reference strain (except the nontoxigenic one). In vitro toxin A was detected by TCD toxin A EIA in 42 strains. These results were compared with those obtained in Tox A/B ELISA test. We observed 52 positive strains. Toxigenic reference strain and two Japanese toxA(-)/toxB(+) strains were also positive. Only 2 negative results were obtained with the nontoxigenic reference strain and unique nontoxigenic isolated strain. Tox A/B ELISA test seems to be the best for detection of C. difficile toxins in vivo and in vitro. Test avoids the false-negative results in the case of presence of toxin A-negative and toxin B-positive strain.