The extracts from the roots of Salvia miltiorrhiza Bunge (Danshen) are widely and traditionally used in the treatment of angina pectoris, acute myocardial infarct, hyperlipidemia and stroke in China and other Asian countries. In this study, we have investigated the role of P-glycoprotein (P-gp) in the intestinal absorption of tanshinone IIA (TSA), a major active constituent of Danshen, using several in vitro and in vivo models. The oral bioavailability of TSA was about 2.9-3.4% in rats, with non-linear pharmacokinetics when its dosage increased. In a single pass rat intestinal perfusion model, the permeability coefficients (Papp) based on TSA disappearance from the luminal perfusates (Plumen) were 6.2- to 7.2-fold higher (P < 0.01) than those based on drug appearance in mesenteric venous blood (Pblood). The Pblood, but not Plumen, was significantly increased when co-perfused with verapamil, or quinidine (both P-gp inhibitors). The uptake and efflux of TSA in confluent Caco-2 cells were significantly altered in the presence of verapamil, quinidine, MK-571, or probenecid. The transport of TSA across Caco-2 monolayers was pH-, temperature- and ATP-dependent. Furthermore, the transport from the apical (AP) to basolateral (BL) side of the Caco-2 monolayers was 3.3- to 8.5-fold lower than that from the BL to AP side, but such a polarized transport was attenuated by co-incubated verapamil or quinidine. A polarized transport was also observed in the control MDCKII cells and more apparent in MDR1-MDCKII monolayers, with the Papp values of TSA in the BL-AP direction being 7- to 9-fold higher in MDR1-MDCKII monolayers than those in the control MDCKII cells. Moreover, TSA significantly inhibited P-gp-mediated transport of digoxin in P-gp-overexpressing membrane vesicles with an IC50 of 2.6 μM, but stimulated vanadate-sensitive P-gp ATPase activity with estimated Km and Vmax values of 10.70 ± 0.69 μM and 67.65 ± 1.31 nmol/min/mg protein, respectively. TSA was extensively metabolized to tanshinone IIB (TSB), and two other oxidative metabolites in rat liver microsomes, but the formation rate of TSB in rat intestinal microsomes was only about 1/10 of that in liver microsomes. These findings indicate that TSA is a substrate and reversing agent for P-gp; and P-gp-mediated efflux of TSA into the gut lumen and the first-pass metabolism contribute to the low oral bioavailability. Further studies are needed to explore the role of other drug transporters and first-pass metabolism in the low bioavailability of TSA.