The discovery of a metallic state and a metal-insulator transition (MIT) in two-dimensional (2D) electron systems challenges one of the most influential paradigms of modern mesoscopic physics, namely, that "there is no true metallic behavior in two dimensions". However, this conclusion was drawn for systems of noninteracting or weakly interacting carriers, while in all 2D systems exhibiting the metal-insulator transition, the interaction energy greatly exceeds all other energy scales. We review the main experimental findings and show that, although significant progress has been achieved in our understanding of the MIT in 2D, many open questions remain.In 2D electron systems, the electrons move in a plane in the presence of a weak random potential. According to the scaling theory of localization of Abrahams et al., 1 these systems lie on the boundary between high and low dimensions insofar as the metal-insulator transition is concerned. The carriers are always strongly localized in one dimension, while in three dimensions, the electronic states can be either localized or extended. In the case of two dimensions the electrons may conduct well at room temperature, but a weak logarithmic increase of the resistance is expected as the temperature is reduced. This is due to the fact that, when scattered from impurities back to their starting point, electron waves interfere constructively with their time reversed paths. Quantum interference becomes increasingly important as the temperature is reduced and leads to localization of the electrons, albeit on a large length scale; this is generally referred to as "weak localization". Indeed, thin metallic films and many of the two-dimensional electron systems fabricated on semiconductor surfaces display the predicted logarithmic increase of resistivity.