The transport of radiolabelled albumin from tissue to blood was measured with an external detection technique in isolated, maximally vasodilated rat skeletal muscles. Initially, rat hindlimbs were perfused with albumin-serum solutions containing [99mTc]albumin for at least 2 h, during which time the tracer accumulated interstitially. The accumulated tracer albumin was then washed out over a period of 1 h, using a tracer-free, otherwise identical, perfusate. The wash-out curve was multi-exponential and the last 30-min period was used to calculate the turnover rate constant (k), which was 7.5 x 10(-4) min-1, (+/- 0.7 x 10(-4), n = 5). Moreover, if albumin was assumed to be distributed homogeneously within the interstitium, with a distribution volume (Vi) of 10 ml 100 g-1, a tissue-to-blood clearance of albumin (ClT-B) of 0.0075 ml min-1 100 g-1 could be calculated. By this approach ClT-B is probably slightly overestimated, but is still only 30% of the clearance from blood to tissue (ClB-T), as determined in several previous studies under similar conditions. Thus, transcapillary passage of albumin is highly asymmetrical, being at least three times greater from blood to tissue than in the opposite direction. This is in agreement with the concept of the capillary walls being composed of two populations of functional pores, where macromolecules are transported from blood to tissue mainly by convection through large pores, even at low filtration rates.