Strain rate has a significant effect on the constitutive behavior of metals like structural steels. Classical material dynamic constitutive models assume the effect of strain rate and those of other influencing factors, e.g. plastic strain, to be uncoupled, but past experiments showed otherwise. In this paper, a series of dynamic tensile tests of Q355 steel was conducted under five strain rates over the range of 10−3s−1 to 5 × 103s−1, using the Instron 5982 universal testing machine, Zwick/Roell HTM5020 high-speed tensile testing machine and split Hopkinson tensile bar (SHTB) system. Full-range true stress-strain relationships corresponding to all strain rates were also determined with the aid of FE simulations. The strain rate effect was found to be dependent on the plastic strain by comparing true stress-strain curves under five different strain rates. Moreover, the effect became less prominent with the increase in plastic strain, in other words, the values of DIF on the true stress decreased gradually from DIFy to DIFu and then leveled off towards the end of plastic strain, leading to the values of DIFy were much higher than DIFu and DIFavg. A new coupled dynamic constitutive model was accordingly proposed by adding a term that reflects the strain rate-plastic strain coupled effect to the Swift-Voce model. The coupled term was achieved by describing the curves of DIF on the true stress versus true strain with an exponential equation, in which, the constants in the equation corresponding to each strain rate were all replaced by a rate-dependent expression. The new coupled constitutive model was proved to provide much more accurate predictions of the constitutive behaviors compared to the traditional uncoupled constitutive models, e.g. J-C and C-S models; in which, the materials used for the comparison included Q355 steel in this paper as well as others collected from the literature (e.g., Q235 steel, Q355 steel and Q460 steel).