Objective To evaluate five detection methods for the laboratory diagnosis of Clostridium difficile infection in the hospitals of USA, and explore a sensitive, specific, accuracy and rapid regimen for the early diagnosis of Clostridium difficile infection. Methods A total of 174 stool specimens submitted to the clinical microbiology laboratory for Clostridium difficile testing were separately tested by five methods including toxigenic culture (TGC), Premier Toxin A&B EIA( A/B-EIA), C. Diff Quick Chek Complete( DEIA), BD G eneOhm Cdiff assay(BD-PCR) and Laboratory-developed PCR(LD-PCR). The gold standard of TGC was used as a reference criterion, and the sensitivity, specificity, positive predictive value ( PPV )and negative predictive value (NPV) of A/B-EIA, D-EIA, BD-PCR and LD-PCR assays were determined. Results Among the 174 specimens studied, 24 were defined as true positives for Clostridium difficile infection by TGC assay, giving a positive rate of 13.8% (24/174). In comparison to the standard,the sensitivity, specificity, PPV and NPV were 62.5%, 99.3%, 93.8% and 94.3% for A/B-EIA;66.7%, 98.7%, 88.9% and 94.9% for D-EIA; 83.3%, 98.7%, 90.9% and 97.4% for BD-PCR;79.2%, 93.3%, 65.5% and 96.6% for LD-PCR. Among all tested specimens, 34 were positive by atleast one of five methods, and of which 15 were concordant by all five methods. The D-EIA results were divided into three groups depending on results of GDH and (or) toxins A/B: 18 were positive for both GDH and toxins A/B, 23 were positive for only GDH, and 133 were negative for both GDH and toxins A/B. Of 18 positive specimens by D-EIA assay, all were concordant with results of BD-PCR assay and 16 were agreement with results of TGC assay. Twenty-two of 24 positive specimens by TGC assay were included in 41 specimens that were positive for GDH. Among eight false negative specimens by D-EIA assay, four were differentiated as positive results by BD-PCR. According to the present study, the sensitivity, specificity,PPV and NPV of a two-step detection algorithm in combination with D-EIA and BD-PCR assays were 83.3%, 98.7%, 90.9% and 97.4%, respectively. Conclusions From the point of technological evaluation, BD-PCR is preferable. A two-step detection algorithm combining D-EIA with BD-PCR is proposed for the laboratory diagnosis of Clostridium difficile infection. This algorithm has demonstrated an excellent sensitivity and specificity, as well as decreased test turnaround time and test cost.