A laboratory investigation of the accuracy and reliability with which tests on models, particularly small-scale models, can be used to predict the behaviour of reinforced concrete structures is described in this paper. The first part of the paper deals with materials which can be used to construct a model that may be expected to give realistic predictions of the behaviour of the full-sized structure, or prototype, at all stages of loading up to and including failure. The second part gives the results of tests on several series of geometrically similar structural elements of various types. The tests showed that there were appreciable scale effects, the smaller specimens giving greater strengths, both in the model concrete, or microconcrete, and in the structural elements themselves. If these factors are properly taken into account, tests on models can provide a workable method of predicting the behaviour of full-sized reinforced concrete structures or of obtaining data which can serve as a basis for mathematical analysis. /TRRL/