A number of instrumental means to predict cooked rice texture has been reviewed. However, Little information has been reported as to a direct comparison for the two typical compression tests (double vs. single) in performance to predict cooked rice texture. This study was aimed at exploring the performance of a double compression (DC) and single compression (SC) test for predicting cooked rice texture and a potential use of Partial Least Square Regression (PLSR) for developing predictive models of specific texture attributes. Four different cultivars of rice stored for 32 weeks were used in this study. A total of 11 texture attributes of cooked rice in five stages were profiled by 7 trained descriptive panelists. Five sensory attributes (manual stickiness, initial cohesion, adhesion to lips, toothpull and hardness) showing significant differences by descriptive panel between rice samples over different storage time were finally predicted. The models by a DC and SC test were robust as well as discriminative and equivalent in performance for predicting texture of cooked rice. Both tests allowed the satisfactory prediction for adhesion to lips and toothpull and the moderate prediction for manual stickiness, initial cohesion and hardness. However, considering that it is routine assessments for rice breeders to predict mechanically rice texture quality, a SC test would have the advantage being less time-consuming over a DC test.