Programmed subcutaneous insulin infusion is a promising method for normalisation of the blood glucose concentration in insulin-dependent diabetics. To evaluate insulin availability the absorption rate from the depot is usually measured intermittently by radioactively-labelled insulin and stationary scintillation detectors. Small portable detectors are an alternative, however, and continuous absorption measurements could be made during normal life conditions. Contrary to conventional single injection therapy, the insulin depot initially expands during infusion treatment, changing the geometry during measurements. In the present study the methodological aspects and geometrical dependences were investigated. Simulated studies were made with various plane disc 125I sources in Perspex phantoms as well as 125I-insulin absorption studies in short-term subcutaneous infusion experiments with anaesthetised rabbits. Results from portable, end-window Geiger-Müller (GM) detectors fixed above the depots and close to the surfaces of phantom or skin were compared with results obtained by a conventional stationary NaI(Tl) detector 15 cm from the phantom or skin surface. With a 125I-insulin infusion site at 5 mm depth in the subcutaneous tissue of rabbits, an overall linear proportionality was found between the results obtained with a NaI(Tl) detector and a GM detector raised 15 mm above the skin surface inside the detector housing.