The accuracy of tensile strengths of concrete obtained by the uniaxial tension test proposed by the authors was investigated, comparing with simulation results which were calculated by using finite element models combined with fictitious cracks and bi-linear tension softening curve. The following were clarified. The application of notches is reasonable procedure, because the missleadings in observing tensile strength of concrete is insignificant in spite of high stress concentration in elastic region due to the presence of notches. It is indispensable to eliminate a secondary flexure which usually occurs during the uniaxial tension test, because the reduction of the observed tensile strength caused by the secondary flexure sometimes reaches 20%. The results show the reliability of the proposed uniaxial tension test.
