Thromboxane B2 (TxB2) determination is usually performed by using commercial 3H-RIA kits. However, the low amounts of TxB2 present in plasma are not detectable without previous extraction. The aim of this study is the evaluation of 1) plasma protein interferences on the binding and separation steps of bound from free analyte and 2) charcoal efficacy in different experimental conditions. Our results indicate that plasma proteins do not influence the antibody binding, but significantly reduce the efficacy of precipitation of kit dextran-charcoal, so that the supernate radioactivity rises with the protein amount increase (r = 0.99 p less than 0.001). Such greater number of counts in the samples determines a lower estimation of TxB2 concentration in plasma when the calibration curve is set up in buffer. Our findings suggest that, in order to measure low amounts of plasma TxB2 without extraction, it is useful: 1) to refer to a calibration curve set up in buffer-diluted plasma, 2) to use the uncoated charcoal concentration allowing the lowest stripping and 3) to perform all steps at 4 degrees C.