This paper critically reviews the main empirical research on goodwill accounting with the purpose of informing and contributing to current debates: the application of a systematic amortisation plus an impairment when required (amortisation model) or an annual impairment-only test (impairment model). Using the main databases (ABI inform, ProQuest Central, Emerald, Science Direct, Scopus and Google Scholar), this critical review highlights the difficulty to resolve doubts at this stage. Arguments for and against the amortisation and impairment models are found. Nevertheless, going back to a systematic amortisation does not seem to be the solution but the impairment test model is eliminated. We also note that there is more room for improvement of the impairment model. Thus, we provide some guidelines and recommendations to improve it. Finally, we find that further investigation can be carried out to fill the gaps identified in the literature and we make recommendations for future research projects.