The rule of mixtures (ROM) method is often used to estimate the tensile strength of fiber reinforced polymers (FRPs) reinforcing bars (rebars). Generally, the ROM method predicts the FRP rebars’ modulus of elasticity adequately but overestimates their tensile strength. This may result from defects occurred during manufacture that prevent the used materials from exhibiting a sound performance and the shear-lag phenomenon by transmission of external forces through the surface of the rebar having a circular cross section. Due to the latter, there is a difference in fiber breaking points regarding the fibers located on the surface and fibers located at the center, and thus results in differences between the values calculated from the conventional ROM and the experimental result. In this study, for the purpose of resolving the problem, glass FRP (GFRP) rebars were shaped to have a hollow section at the center of their cross sections and were further subject to tensile strength tests. The test results were further placed under regression analysis and a modified ROM within ±5% accuracy compared to the experimental value was proposed for GFRP rebars with 13, 16, and 19 mm diameters.