Methods of modelling power systems in the study of the system transient behaviour are reviewed. Six machine models of varying complexity are developed from the fundamental machine equations. Factors in the form of the machine equations, which critically influence the economy of solution, are identified, The relationship between the machine and system equations is examined in relation to the digital solution of multimachine problems, and generalised methods of solution of the equations are outlined. Comparisons are made between results obtained from the range of machine models and those obtained from full-scale fault-throwing tests on the public supply system. In one case, a single machine connected to a system of high capacity is modelled, while, in the second example, two 60 MW machines connected to a large mulrimachine system is modelled. Both non-pole-slipping and pole-slipping tests for the case of the 2-machine system are examined, in which a machine marginally remained in synchronism and marginally lost synchronism, respectively. It is shown that models in which subtransient phenomena are simulated, but in which some transformer voltages in the stator equations, together with subtrasient saliency, are neglected, provide an adequate and economical model for system disturbances, including those in which the machine may fall from synchronism.