This study aimed to further the understanding of the incidence of adverse events (AEs) in a population-based representative liver cancer population where there is currently a lack of knowledge. We carried out a retrospective cohort study using data from an administrative claims database between 1 January 2004 and 31 December 2010. Patients were included in the study if they had at least one primary liver cancer diagnosis [International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM): 155.0] and a metastatic diagnosis [ICD-9-CM: 196.x, 197.x (except 197.7), 198.x or 199.0]. We estimated the incidence rate (IR) and 95% confidence interval (CI) for each AE under study. Of the patients identified, 1292 fulfilled the inclusion and exclusion criteria. The most common AEs were nausea and vomiting (IR=878.5/1000 person-years; 95% CI=799.5-963.1). Other common AEs were hypertension (IR=648.7/1000 person-years; 95% CI=569.2-736.1) and hemorrhage (IR=580.0/1000 person-years; 95% CI=518.6-646.6). The least common AEs were rare dermatologic diseases such as Stevens-Johnson syndrome and toxic epidermal necrolysis where no cases were observed. The rates detailed in this analysis are helpful in understanding the benefit risk of treating patients with liver cancer in the real world. Although no formal comparisons were performed, the increased risk of certain events observed in sorafenib-treated patients from this analysis mirrors the risks reported on the label for sorafenib. Therefore, this analysis provided a reasonable assessment of the AEs that patients with liver cancer experience in the real world.