Due to the difficulties inherent in the direct measurement of concrete tensile strength, experimental methods have been devised to determine this property indirectly. One such method, the splitting or indirect tensile test, also known as the Brazilian test, is widely used for its convenience and applicability to freshly moulded cubic and cylindrical specimens, as well as to hardened concrete core samples.Nevertheless, the strength values obtained with this method have been observed to differ with specimen size. Moreover, the strength values calculated with the formula proposed in the existing standards fail to reflect real test conditions and actually overestimate tensile strength.The experimental results reported in this article show that if the test set up is modified slightly and the equation for calculating strength is adjusted, consistent strength values are obtained, regardless of specimen size, in the range of diameters studied (45 - 294 mm).