The aim of the present systematic review was to evaluate the available evidence on laboratory diagnosis of CDI and to formulate recommendations to optimize CDI testing. In comparison with cell culture cytotoxicity assay (CCA) and toxigenic culture (TC) of stools, we analyzed the test characteristics of 13 commercial available enzyme immunoasssays (EIA) detecting toxins A and/or B, 4 EIAs detecting Clostridium difficile glutamate dehydrogenase (GDH), and a real-time PCR for C. difficile toxin B gene. In comparison with CCA and TCA and assuming a prevalence of CDI of 5%, PPV and NPV varied between 0.28-0.77, 0.12-0.65 and 0.98-1.00, 0.97-1.00, respectively. Only if the tests were performed in a population with a CDI prevalence of 50 percent, would PPVs be acceptable (ranging from 0.71 to 1.00).To overcome the problem of a low PPV, we propose a two step approach, with a second test or a reference method in case of a positive first test. Further reducing the number of false negative results would require either retesting of all subjects with a negative first test, or re-testing all subjects with a negative second test, after an initially positive test. This approach resulted in non-significant improvements, and emphasizes the need for better diagnostic tests. Further studies to validate the applicability of two-step testing, including assessment of clinical features, are required.