Due to the assumption of uniform bond stress, the development length of fibre reinforced polymer (FRP) bars by design standards can be unnecessarily long and difficult to provide in practice. Hence, the bond stress distribution and required development length of a glass fibre reinforced polymer (GFRP) rebar is investigated. Four beam-bond specimens, two following RILEM specifications and two based on a procedure by ACI are tested to evaluate the effect of test method on bond strength. Ten pullout tests are also performed using the same bar. The two test methods yield similar results, but the ACI test is easier to perform. The bond stress distribution in the beams is highly nonlinear but in the pullout tests approaches uniformity. The actual development length is found to be 50% to 250% less than that required by the aforementioned standards. Consequently, a new equation is proposed based on the logistic growth function to model the non-uniform bond stress distribution and estimate the required development length.