Abstract: As a necessary input into the finite-element models aimed to evaluate the plastic response and ultimate resistance of steel structures, the full-range true stress-true strain relationship of steel material should be accurate to ensure reliable simulation results. Due to the occurrence of necking in the later stage of tensile coupon test, which results in localised plastic strain, the post-necking true stress-true strain relationship cannot be computed from the engineering stress-engineering strain relationship directly. This paper proposes a new model, i.e. the Generalised Post Necking (GPN) model, to obtain the accurate post-necking true stress-true strain relationship. The GPN model has the following advantages: the true stress-true strain curve predicted by the GPN model is differentiable at the necking point; the reduction in the post-necking strain hardening modulus can be adjusted in a controlled way; and the GPN model is applicable to various steel grades (LYP160 low yield steel to 12.9 grade high-strength bolt). More importantly, it has been verified by results of material coupon tests and steel connection tests that the GPN model has higher accuracy than traditional models such as the Swift-Voce model, Ling model and Modified Weighted Average (MWA) model. In addition, through analysis of 76 engineering stress-engineering strain curves covering various steel grades, the features of the post-necking engineering stress-engineering strain curves are revealed, and the methods for characterising different post-necking engineering stress-engineering strain curves are proposed. The correlation between the post-necking true stress-true strain curve and engineering stress-engineering strain curve is established for the GPN model parameters to be directly computed without the need of complementary finite-element simulations.