Liver cirrhosis is the ultimate consequence of the wound healing reaction subsequent to a chronic injury, which leads to a complete derangement of the normal hepatic lobular and vascular architecture. Cirrhosis is characterized by patterns of evolution depending on the causative agent and a series of complex underlining mechanisms in which neo-angiogenesis and necro-inflammation play a key role. The importance of the different cell types involved and of the extracellular matrix composition as well as the role of innate immunity, bacterial translocation and oxidative stress are also emerging. A variable degree of regression of fibrosis and even cirrhosis has been described, in experimental models, after suspension of the liver disease causative agent. As some individual features influence the rate of fibrosis progression, genetic and epigenetic factors are likely to influence fibrosis regression. Key Messages: There is increasing awareness that cirrhosis is not a static condition but a dynamic process. Current semi-quantitative scores and clinical classifications are inaccurate and unable to identify the different phases of evolution of the advanced stages of chronic liver diseases (CLDs). The increasing availability of effective etiology-driven therapeutic options for CLDs makes reversion of cirrhosis a more possible prospective. However, the removal of the causing agent, depending on the stage of the disease, does not necessarily eliminate the risk of disease progression, decompensation and development of hepatocellular carcinoma. Also, the non-invasive markers currently validated for the assessment of fibrosis are not suitable for an effective evaluation of fibrosis regression. Conclusions: There is a critical need of a system that would be able to more accurately describe the dynamic development of cirrhosis and the impact of tissue fibrosis, neo-angiogenesis, necro-inflammation and attempted regeneration on its evolution. Effective treatment of CLD can lead to a variable degree of fibrosis regression. New markers able to evaluate this process will need to be detected and validated.