Non-alcoholic steatohepatitis is a disease without a single, specific, diagnostic marker, hence multiple indicators are required to measure therapeutic efficacy. Moreover, drug candidates for non-alcoholic steatohepatitis target many distinct mechanisms that are believed to promote hepatic injury. Therefore, a wide range of endpoints must be reached, sequentially, as required by the drug development process. Some of these endpoints validate the mechanism of action, others are used to anticipate histological efficacy. Histological endpoints are still considered the best predictors of clinical outcome, but they can only be reliably tested in larger, late phase trials. Herein, we will review the rationale and clinical data supporting the use of specific endpoints at different stages of therapeutic trials. We will also discuss the validity and limitations of current phase IIb histological endpoints, particularly a one stage reduction in fibrosis, for their ability to predict progression to cirrhosis, which is the ultimate outcome measure in therapeutic trials.