Abstract : Evidence-based medicine (EBM) is mainly supported by the results of randomised controlled trials (RCTs). If the latter offer guarantees of reliability, especially by minimizing the influence of confounding factors and potential biases, they also have limitations. Observational databases resulting from real life registries, if possible build in a prospective manner, may offer some solutions, but are also exposed to limitations. This article compares the advantages and disadvantages of the two sources of information, which ideally should be complementary. For the purpose of illustration, we shall compare the recent results of RCTs and of observational databases from multinational registries that investigated the effects of sodium-glucose cotransporter type 2 inhibitors (gliflozins) on cardiovascular outcomes in patients with type 2 diabetes.