Aim: The ability to return interstitial protein to central blood is key to the defence against oedema. The aim of this study was to quantify this ability by measuring the rate at which radiolabelled human immunoglobulin (HIgG) accumulated in blood following injection into the subcutis of the hand in normal volunteers and in patients with breast cancer-related lymphoedema (BCRL). Methods: A total of 37 control subjects (healthy normal volunteers or breast cancer patients prior to treatment) and 18 women with BCRL were studied with dual-isotope lymphoscintigraphy. Each received bilateral subcutaneous depot injection in the dorsal web space of HIgG labelled with Tc-99m on one side and In-111 on the other. Activities remaining at the depot and accumulating in blood were measured at regular intervals for 3 h. Clearance from the depot was exponential and expressed as the rate constant k<sub>depot</sub> (min<sup>–1</sup>). Accumulation in blood was essentially linear and, using an estimate of blood volume based on height and weight, was expressed as the linear constant b<sub>blood</sub> (% administered activity·min<sup>–1</sup>). The time axis intercept of this linear fit was recorded as an index of the minimum time to arrival of radioprotein in blood. The efficiency with which radioprotein that has left the depot (extra-depot activity) is transported into blood [tissue-to-blood (T-B) transport] was quantified (1) as the quotient b<sub>blood</sub>/k<sub>depot</sub>, and (2) as a function of time after injection by comparing the total amount of radioprotein in blood at any time with the total amount of radioprotein that was no longer in the depot at the same time. Results: Tc-99m-HIgG and In-111-HIgG behaved similarly and are interchangeable. At all times between 60 and 180 min in controls, about 50% of protein that had left the depot was present in blood. T-B transport was reduced to about 20% in BCRL arms in which the hand was involved in swelling (p < 0.001 versus controls), but remained unchanged in patients in whom the hand was spared. The minimum time to arrival of radioprotein in blood was not reduced in BCRL; on the contrary, there appeared to be a small proportion of injected activity that arrived rapidly in blood in BCRL patients but not in controls. Conclusion: We conclude that T-B transport is only impaired in BCRL when radioprotein is injected into swollen tissue. Significant quantities of radioprotein may escape from the arm via local access to blood. Individual variation in this capacity may explain the regional sparing observed in BCRL.