The plasma protein binding and distribution in blood cells of the novel anti-tumour agent 5,6-dimethylxanthenone-4-acetic acid (DMXAA) has been investigated in-vitro using filtration and an HPLC method to measure DMXAA. DMXAA (500 microM) was extensively bound in plasma from all species with an unbound fraction (fu) of 4.61+/-1.10 (mouse), 2.59+/-0.32 (rat), 2.02+/-0.48 (rabbit) and 2.07+/-0.23% (human). The binding was concentration dependent with DMXAA concentrations > or = 1,000 microM markedly increasing the fu in the plasma from all species. The estimated number of binding sites in plasma were 2.4+/-0.2 (mouse), 1.7+/-0.2 (rat), 0.8+/-0.1 (rabbit) and 2.1+/- 0.2 (human). The major binding protein in human plasma was albumin, with negligible binding to gamma-globulin and alpha1-acid glycoprotein. There was a significant linear relationship between the bound:free DMXAA concentration ratio (Cb/Cu) and albumin concentration in human serum albumin solution (r = 0.955; P < 0.05) and in healthy human plasma (r = 0.998; P< 0.05), but not in plasma from cancer patients (n = 5), nor across species. In cancer patients (n = 5) DMXAA had a significantly higher (P < 0.05) fu (4.60+/- 0.42%) compared with healthy human plasma (2.07+/-0.23%). In human plasma, the fu of DMXAA (500 microM) was significantly reduced by 500 microM diazepam (P < 0.05), but not by warfarin, phenylbutazone, salicylic acid, ibuprofen or clofibric acid at that concentration. DMXAA significantly reduced the binding of dansylsarcosine (a Site-II binder) to HSA, but significantly increased the binding of dansylamide (a Site-I binder). Within species, the blood:plasma concentration ratio (CBL/CP) of DMXAA was relatively constant (mouse, 0.581+/-0.005; rat, 0.667+/-0.025; rabbit, 0.637+/-0.019; human, 0.673+/-0.103) over the range 50-1000 microM, but increased significantly at DMXAA concentrations > 1000 microM in all species except the rabbit. These results indicate that significant alterations in DMXAA plasma binding and distribution into blood cells occur with increasing concentrations of DMXAA in all species, and also that significant interspecies differences exist. It would be more appropriate to compare plasma unbound concentrations when assessing DMXAA exposure in cancer patients or when extrapolating across species.