Selective internal radiation therapy (SIRT) of hepatocellular carcinoma (HCC) has been used for many years, usually without any specific dosimetry endpoint. Despite good clinical results in early phase studies or in cohort studies, three randomized trials in locally advanced HCC available failed to demonstrate any improvement of overall overall survival (OS) in comparison with sorafenib. In recent years, many studies have evaluated the dosimetry of SIRT using either a simulation-based dosimetry (macroaggregated albumin (MAA)-based) or a post-therapy-based one (90Y-based). The goal of this review is to present the dosimetry concept, tools available, its limitations, and main clinical results described for HCC patients treated with 90Y-loaded resin or glass microspheres. With MAA-based dosimetry, the threshold tumor doses allowing for a response were between 100 and 210 Gy for resin microspheres and between 205 and 257 Gy for glass microspheres. The significant impact of the tumor dose on OS was reported with both devices. The correlation between 90Y-based dosimetry and response was also reported. Regarding the safety, preliminary results are available for both products but with a larger range of normal liver doses values correlated with liver toxicities due to numerous confounding factors. Based on those results, international expert group recommendations for personalized dosimetry have been provided for both devices. The clinical impact of personalized dosimetry has been recently confirmed in a multicenter randomized study demonstrating a doubling of the response rate and an OS of 150% while using personalized dosimetry. Even if technical dosimetry improvements are still under investigation, the use of personalized dosimetry has to be generalized for both clinical practice and trial design.