Experimental evidence has accumulated over the years to show that a variety of strategies that limit neovascularisation also slows down or inhibits tumour growth, suggesting that blocking tumour-induced angiogenesis is a valid, novel approach to tumour therapy (Folkmann 1997).The rat Subcutaneous Air Sac (SAS) model (Lichtenberg et al. 1997(Lichtenberg et al. & 1999 was recently described as a new in vivo model designed to replace a currently used eye model where neovascularisation is induced by chemical injury to the cornea of the rabbit or by implantation of tumour cells intracorneally (Gimbrone et al. 1974). In the SAS model the membrane of the subcutaneous air sac simulates the cornea as a macroscopically avascular substrate in which the new vessels can develop. The SAS model has shown high reproducibility and is considered more ethically acceptable. However, limitations of the test system are related to the timeconsuming phase of membrane induction (10-14 days) and that the model has been validated for rats only. Often the mouse is preferred in the screening phase obviously because less material is required for testing. Attempts to adapt the SAS technique to the mouse have not been successful. It is the experience of the authors that a functional subcutaneous membrane similar to the one obtained in rats cannot be established in the mouse. In rats the cell layer of the membrane is of fibrous stromal structure containing 4-10 layers of cells (Ingeborg et al. 1986). In mice the membrane is assumed to be thinner and not capable of eliciting a clear angiogenic response. Here is presented a methodology in mice closely related to the rat subcutaneous air sac model. In this mouse model the time-consuming induction of a subcutaneous membrane is obviated and replaced by a subcutaneous sponge implant. In the centre of the sponge is placed a strong mitogenic growth factor, basic fibroblast growth factor (bFGF), known to induce new vessel growth.Thirty-eight NMRI female mice (SPF Bom:NMRI)Author for correspondence: J. Lichtenberg, Department of Preclinical Development, NeuroSearch A/S, Pederstrupvej 93, 2750 Ballerup, Denmark (e-mail jel/neurosearch.dk).weighing 25-27 g were obtained from M&B, Ejby, Lille Skensved, Denmark. They were housed in a facility where light was controlled on a 12 h light-dark cycle. The room temperature and relative humidity recorded by a thermohygrograph showed values between 20.5-24.1 ae and 40-67%, respectively. The animals were fed a pelleted rodent diet (Altromin 1324, Brogården, Denmark) ad libitum and had free access to tap water. All animals were observed daily for clinical signs. Slow-release pellets containing 200 ng of human basic fibroblast growth factor (Innovative Research of America, FL, USA) were circular and with a diameter of 1.5 mm. The angiogenic peptide was guarantied by the supplier to be released over a period of 10 days.The mice were anaesthetized using inhalation anaesthesia (halothane/N 2 O and oxygen). The skin of the back was