Abstract: In this paper, creep deformation and recovery tests were processed for tomato fruits first with a TA-XT2i texture analyzer by randomly selecting every ten tomato fruits for a different ripening period. It was found the classical Burger's model failed to express the strain saturation at the later creep objectively due to the constant Newtonian viscosity of the free dashpot, but the existing modified exponential and power-law models have also have the major defect that the viscosity of free dashpot at the later creep and initial deformation rate tend to infinity, respectively. So a new model that consisted of four elements with six parameters modified was proposed, by changing the constant Newtonian viscosity of a free dashpot to a power-law variable with a constant term in the viscous deformation term of the creep equation of Burger's model. And then the curve fitting and prediction results of the new model were compared with those of Burger's model and the existing modified exponential and power-law models. The average determination coefficient and the sum of squares due to the regression error of different ripening periods with the new model were 0.9975-0.9994 and 0.04047-0.07633, respectively, which are more ideal, obviously. The average relative errors of the rate of creep deformation, and the rate of elasticity with the new model were only 2.54%-3.04% and 2.89%-7.12%, respectively. The average relative error of the creep deformation prediction with the new model were 0.29%-0.46%, which are several times less than not only the Burger's model but also the existing modified exponential and power-law models. Finally, the initial creep deformation rate of the new model was also closer to the real rate than any other models. To conclude, this study used a good 4-element with 6-parameter modified model to express and predict the creep behavior more accurately, which gives a better model option to bruise reducing in fruit harvesting, transporting or sorting.