We evaluated two rapid toxin tests for C. difficile combined with stool specimen cultures used from January 2006 to March 2009. Stool specimens numbered 877, 102 among which were from the cases of diagnosed clinical C. difficile-associated diarrhea (CDAD). Rapid toxin A 'Uniquick' detection kits were used until October 2007 and toxin A&B 'TOX A/B' detection kits thereafter. Clinical CDAD was considered the detection gold standard. Uniquick sensitivity, specificity, and positive and negative predictive values were 54.3%, 99.1%, 90.5%, and 93.2% while those for TOX A/B were 46.2%, 97.6%, 65.2%, and 95.0% and for culture 42.2%, 95.5%, 55.1%, and 92.6%. Rapid toxin tests tended to have better sensitivity than culture results although not significantly so, and Uniquick showed significantly better positive predictive value than TOX A/B or culture results. Among clinical CDAD cases, concordance with culture was 24.3% for Uniquick and 53.1% for TOX A/B. For stored strains, 27 were typed toxin A+B+ (48.1%), toxin A-B+ (37.0%) and toxin A-B- (14.8%) with toxin gene detection by PCR. Eight of the 10 toxin A-B+ strains were classified into two cluster by ribotyping, and 7 of those were detected in two hospital wards, indicated the possibility of nosocomial toxin A-B+ strain spread. The rapid toxin test for both toxins A and B should be used if toxin A-B+ predominate. Simultaneous culture testing may be useful for detecting clinical CDAD more accurately, however.