Among the mechanical parameters that are used to describe the quasi-elastic properties of bituminous mixtures, special attention should be devoted to the so-called elastic stiffness which can be measured by means of the repeated load indirect tensile (RLIT) test. As repeatedly reported in literature such a parameter can be successfully used in experimental design methods and for the estimation of the comparative field performance of different mixtures. As a result of its continuous use in the Road Materials Laboratory (RML) of the Politecnico di Torino, the Authors have thoroughly assessed in the past three years the enormous potential of the RLIT test, also highlighting its weaknesses and flaws. In this paper two important issues are addressed. In first place an evaluation of the random and systematic errors associated to RLIT test measurements has been carried out by means of an adequate mathematical approach. Moreover, the effects of a number of test variables have been modelled and described in order to enhance the significance of test results in the general framework of performance-related mix design and quality control. These analyses have lead to the definition of a new and improved testing protocol for the RLIT test which is capable of guaranteeing a grater stability of the results, and to the suggestion of possible improvements to the currently used testing apparatus.