We present a technique to reconstruct the surface of the surgical field during neurosurgery. The images are taken with a stereo camera mounted on a surgical microscope, which is part of a neuronavigation system. We use an intensity-based method to calculate a dense disparity map over the whole image extend. Correspondences are established by exploiting the epipolar ge- ometry and comparing image intensities along corresponding epipolar lines. The method does not require a strong calibration. Only the colineation denoted by the fundamental matrix, which maps corresponding epipolar lines from one image to the other, has to be known. The results show that our method is able to reconstruct the surface of the surgical field upto a few areas. It is limited at points of specular reflections of the wet brain surface and points of uniform texture.