This study dealt with a new instrumental method for grading fabric softness. A pure cotton plain weave and 35/65 cotton/polyester twill were washed with Tide ^® and softened with Downy ^® . Variations of fabric mechanical properties determined by the Downy ^® treatment was measured using an LSU instrument composed of a QT/5 tensile tester, a set of attachments, and test methods software. The neural network software NeuroSolutions was employed in predicting models for grading fabric softness based on the measured fabric mechanical properties. To verify the grading models, the softener Snuggle ^® was used as a test softener. There was no predicting error in the present study. The new method of fabric softness evaluation is suitable for practice in a dynamic industrial environment because the new instrument can be routine laboratory equipment and the neural networking models can be updated on a daily basis to include new types of fabrics and softeners.