In this paper we present a new dynamical systems algorithm for clustering in hyperspectral images. The main idea of the algorithm is that data points are \`pushed\' in the direction of increasing density and groups of pixels that end up in the same dense regions belong to the same class. This is essentially a numerical solution of the differential equation defined by the gradient of the density of data points on the data manifold. The number of classes is automated and the resulting clustering can be extremely accurate. In addition to providing a accurate clustering, this algorithm presents a new tool for understanding hyperspectral data in high dimensions. We evaluate the algorithm on the Urban (Available at www.tec.ary.mil/Hypercube/) scene comparing performance against the k-means algorithm using pre-identified classes of materials as ground truth.