Abstract. Recent progresses in the description of glassy relaxation and ageing are reviewed for the wide class of network-forming materials such as GeO 2 , Ge x Se 1−x , silicates (SiO 2 -Na 2 O) or borates (B 2 O 3 -Li 2 O), all of them having an important usefulness in domestic, geological or optoelectronic applications. A brief introduction of the glass transition phenomenology is given, together with the salient features that are revealed both from theory and experiments. Standard experimental methods used for the characterization of the slowing down of the dynamics are reviewed. We then discuss the important role played by aspect of network topology and rigidity for the understanding of the relaxation of the glass transition, while also permitting analytical predictions of glass properties from simple and insightful models based on the network structure. We also emphasize the great utility of computer simulations which probe the dynamics at the molecular level, and permit to calculate various structure-related functions in connection with glassy relaxation and the physics of ageing which reveals the off-equilibrium nature of glasses. We discuss the notion of spatial variations of structure which leads to the picture of "dynamic heterogeneities", and recent results of this important topic for network glasses are also reviewed.PACS numbers: 61.43. Fs, 64.70.kj Relaxation and physical ageing in network glasses 2 Figure 1. Typical network-forming glasses: a) A stoichiometric glass former (SiO 2 , B 2 S 3 ) whose structure and network connectivity can be altered by the addition (b) of 2-fold coordinated atoms (usually chalcogens, S, Se) that lead to cross-linked chains. The structure can also be depolymerized (c) by the addition of a network modifier (alkali oxides or chalcogenides, Na 2 O, Li 2 S, etc.). Glassy dynamics depends strongly on the network topology, i.e. the way bonds and angles arrange to lead to a connected atomic network. Note that only chalcogenides can produce a mixture of these three kinds of basic networks, e.g. (1-x)Ge y Se 1−y -xAg 2 Se [2], and for the latter system x=0 corresponds to case (b), y=33% corresponds to case (c), and both conditions together (x=0,y=33 %) to case (a).