Historically, the liver was thought to be an organ unsuitable for radical doses needed to treat primary or secondary tumors. This determination was made in the time where only 2D treatment planning was available (Ingold et al. 1965; Wharton et al. 1973). However, the consequent interpretation of liver toxicities with 3D-conformal radiotherapy with the use of dose–volume histograms allowed us to describe the normal tissue complication probability (NTCP) characteristics of the liver, an organ with a parallel tissue structure which is reflected in a high “volume effect parameter,” n = 0.69 (Jackson et al. 1995). The data underlying the NTCP modeling came from a series of 79 patients including nine patients that developed clinical radiation hepatitis. All of the patients with radiation hepatitis, also called radiation-induced liver disease (RILD), had whole-liver radiation with doses of at least 37 Gy in conventional fractionation. On the other hand, patients who had partial liver radiotherapy to much higher doses did not develop RILD. Subsequently, a phase I trial of escalated focal liver radiation and concurrent hepatic artery fluorodeoxyuridine (FUdR) was conducted for patients with unresectable intrahepatic malignancies (Dawson et al. 2000). Twenty-seven patients had hepatobiliary cancer and 16 colorectal liver metastases. This trial employed a dose per fraction of 1.5 Gy twice daily with concomitant intra-arterial FUdR during the first 4 weeks of radiotherapy. Continuous-infusion FUdR required placement of a percutaneous brachial artery catheter to deliver a dose of 0.2 mg/kg/d. The trial was designed to be isotoxic and to escalate radiation dose in cases where the target volumes were small enough to allow dose escalation according to the above-described NTCP model. This resulted in a median radiotherapy dose of 58.5 Gy with a range from 28.5 to 90 Gy. Of note, the median tumor size was as large as 10 × 10 × 8 cm. The dose to the stomach and duodenum was restricted to a maximum of 68 Gy in 1.5 Gy fractions. Twenty-five patients were assessable for response evaluation achieving 16 partial and 1 complete response. Intriguingly, improved progression-free and overall survival depended on multivariate analysis on escalated dose. There was only one incidence of late liver toxicity, namely one patient suffering a reversible grade 3 RILD.