The book is devoted to the theory of gradient flows in the general framework of metric spaces, and in the more specific setting of the space of probability measures, which provide a surprising link between optimal transportation theory and many evolutionary PDE's related to (non)linear diffusion. The book originates from lectures by L. Ambrosio at the ETH Zürich in Fall 2001. It contains new results that have never appeared elsewhere. The book has been substantially extended and revised in cooperation with the co-authors. Particular emphasis is given to the convergence of the implicit time discretization method and to the error estimates for this discretization, extending the well established theory in Hilbert spaces. The book is split in two main parts that can be read independently of each other.