All currently avalable concrete creep predicting models cannot describe well the creep of a concrete structure because they all leave the effects of additives out of consideration.The purpose of this work was to modify model B3 for overcoming this deficiency.We tested thirteen specimens of C40 and C50 concrete with additives,out of which nine were for creep and four for shrinkage tests over a 700 d period under controlled temperature.We compared the experimental results for creep and shrinkage with those obtained by using model B3,and derived modification terms through regression analysis.Based on the experimental results of identical specimens under varied stress levels,we also derived a function considering the effect of stress level on creep.It is suggested that the creep prediction models without adjustment should not be used for modern concrete with a variety of additives. All currently avalable concrete creep predicting models can not describe well the creep of a concrete structure because they all leave the effects of additives out of consideration the purpose of this work was to modify model B3 for overcoming this deficiency.We tested thirteen specimens of C40 and C50 concrete with additives, out of which nine were for creep and four for shrinkage tests over a 700 d period under controlled temperature. We compared the experimental results for creep and shrinkage with those obtained by using model B3, and derived control terms through regression analysis . Based on the experimental results of identical specimens under varied stress levels, we also derived a function considering the effect of stress level on creep. It is suggested that the creep prediction models without adjustment should not be used for modern concrete with a variety of additives .