Background Anthraquinones (aloe-emodin, rhein, emodin, chrysophanol, physcion) and their glucosides are major polyphenol constituents in Rhubarb. An in vivo study revealed anthraquinone glucuronides as the predominant forms in rat plasma when a Rhubarb extract was orally administered, implying impact of glucuronidation on in vivo fates and consequent actions of anthraquinones. Rhubarb anthraquinones share the core structure of danthron (1,8-dihydroxyanthracene-9,10-dione) with different substituents at beta-position (C-3 and/or C-6). The present study investigated, for the first time, the in vitro glucuronidation of rhubarb anthraquinones to determine the structure-metabolism relationship. Methods Aloe-emodin, emodin, chrysophanol, physcion and danthron (0.5 and 5 μM) were individually incubated with pooled hepatic microsomal proteins (0.05 mg/ml) from humans (HLMs) or rats (RLMs) for 10 min in the presence of UDPGA. The incubation of rhein was 30 min and 0.5 mg/ml of proteins. In incubations with recombinant human UGT (0.05 mg/ml, 12 isozymes), four anthraquinones except for rhein were incubated for 10 min. Main metabolites of aloe-emodin and emodin were prepared from scale-up reactions with RLMs, isolated using pre-HPLC, and structures identified using LC-MS/MS and NMR analysis. Chrysophanol-8-O-glucoside and physcion-8-O-glucoside were used to aid structural identification of chrysophanol and physcion glucuronides by glucuronidation and hydrolysis. Quantitative determination was performed using HPLC-MS/MS. Results In microsomal proteins of both species, danthron only produced one glucuronide due to its structural symmetry. All other anthraquinones formed 1-O-glucuronide and 8-O-glucuronide. One additional glucuronide resulting from conjugation of the 3-OH or 3-COOH was identified with emodin and rhein, respectively. RLMs showed higher activities towards glucuronidation of the same anthraquinone than human with exception of emodin, the elimination rate of which was comparable between species. The average elimination rates of test anthraquinones in HLMs and RLMs were in the same descending order (rat vs human, nmol/min/mg protein): emodin (6.80 vs 6.83), aloe-emodin (1.39 vs 0.56), danthron (1.24 vs 0.46), chrysophanol (0.69 vs 0.14), physcion (0.36 vs 0.054) and rhein (0.035 vs 0.016). Substituent(s) at beta-position(s) of anthraquinones also affected positional preference of glucuronidation. In HLMs when substituted with –CH2OH (aloe-emodin) or –CH3 (physcion, -OCH3 at ring B) at ring A, both compounds showed a preference to –OH at ring B. The hydroxyl group at the same benzene ring with –CH3 of chrysophanol tended to be the preferential site of conjugation. With a beta-OH, emodin showed an absolute positional preference with the 3-O-glucuronide accounting for >95% of emodin glucuronides. Aloe-emodin and rhein showed different preference of glucuronidation in RLMs. Human UGT1A1, 1A3, 1A7, 1A8 (except for aloe-emodin), 1A9 and 1A10 catalyzed glucuronidation of all anthraquinones at varied activities and positional preference. UGT 1A9 exhibited the highest activities in most cases. While UGT 2B subfamily showed low or no activities towards antranquinones glucuronidation, except that 2B7 and 2B15 exhibited similar activities to 1A3 towards emodin glucuronidation. All UGTs showed highest capability and predominant preference towards beta-OH of emodin. Conclusions Rhubarb anthranquinones, except for rhein, undergo extensive glucuronidation in liver microsomes of humans and rats. The elimination rate and positional preference of glucuronidation varied with substituted position and moiety.