To identify dosimetric predictors of hepatobiliary (HB) toxicity associated with stereotactic body radiation therapy (SBRT) for liver tumors. We retrospectively reviewed 96 patients treated with SBRT for primary (53%) or metastatic (47%) liver tumors between March 2006 and November 2013. The central HB tract (cHBT) was defined by a 15-mm expansion of the portal vein from the splenic confluence to the first bifurcation of left and right portal veins. Patients were censored for toxicity upon local progression or additional liver-directed therapy. HB toxicities were graded according to Common Terminology Criteria for Adverse Events version 4.0. To compare different SBRT fractionations, doses were converted to biologically effective doses (BED) by using the standard linear quadratic model α/β=10 (BED10). Medianfollow-up was 12.7months after SBRT. Median BED10 was 85.5Gy (range: 37.5-151.2). The median number of fractions was 5 (range: 1-5), with 51 patients (53.1%) receiving 5 fractions and 29 patients (30.2%) receiving 3 fractions. In total, there were 23 (24.0%) grade 2+ and 18 (18.8%) grade 3+HB toxicities. Nondosimetric factors predictive of grade 3+HB toxicity included cholangiocarcinoma (CCA) histology (P<.0001), primary liver tumor (P=.0087), and biliary stent (P<.0001). Dosimetric parameters most predictive of grade 3+HB toxicity were volume receiving above BED10 of 72 Gy (VBED1072)≥21cm(3) (relative risk [RR]: 11.6, P<.0001), VBED1066≥24cm(3) (RR: 10.5, P<.0001), and mean BED10 (DmeanBED10) cHBT ≥14Gy (RR: 9.2, P<.0001), with VBED1072 and VBED1066 corresponding to V40 and V37.7 for 5 fractions and V33.8 and V32.0 for 3 fractions, respectively. VBED1072≥21cm(3), VBED1066≥24cm(3), and DmeanBED10 cHBT ≥14Gy were consistently predictive of grade 3+toxicity on multivariate analysis. VBED1072, VBED1066, and DmeanBED10 to cHBT are associated with HB toxicity. We suggest VBED1072<21cm(3) (5-fraction: V40<21cm(3); 3-fraction: V33.8<21cm(3)), VBED1066<24cm(3) (5-fraction: V37.7<24cm(3); 3-fraction: V32<24cm(3)) as potential dose constraints for the cHBT when clinically indicated.