Experimental exploration of equipment for stereotactic functional neurosurgery based on heating induced by radio-frequency current is most often carried out prior to surgery in order to secure a correct function of the equipment. The experiments are normally conducted in an experimental model including an albumin solution in which the treatment electrode is submerged, followed by a heating session during which a protein clot is generated around the electrode tip. The clot is believed to reflect the lesion generated in the brain during treatment. It is thereby presupposed that both the thermal and electric properties of the model are similar to brain tissue. This study investigates the presence of convective movements in the albumin solution using laser Doppler velocimetry. The result clearly shows that convective movements that depend on the time dependent heating characteristics of the equipment arise in the solution upon heating. The convective movements detected show a clear discrepancy compared with the in vivo situation that the experimental model tries to mimic; both the velocity (maximum velocity of about ) and mass flux are greater in this experimental setting. Furthermore the flow geometry is completely different since only a small fraction of the tissue surrounding the electrode in vivo consists of moving blood, whereas the entire surrounding given by the albumin solution in the experimental model is moving. Earlier investigations by our group (Eriksson, 1999, Med. Biol. Eng. Comput.37, pp. 737–741; Wren, 2001, Ph.D. thesis; and Wren, 2001, Med. Biol. Eng. Comput.39, pp. 255–262) indicate that the heat flux is an essential parameter for the lesion growth and final size, and that presence of convective movements in the model might substantially increase the heat flux. Thus, convective movements of the magnitude presented here will very likely underestimate the size of the brain lesion, a finding that definitely should be taken into consideration when using the model prior to patient treatment.