A diffusion-type time-temperature indicator (TTI) is known to have relatively low temperature dependency, which is an inherent weakness that causes an inevitable prediction error of food shelf-life: under-or over-prediction. In this study, a TTI design method was developed to avoid the undesired overprediction. A polydiacetylene/silica nanocomposite (PDA/SiO2 NC) was used as the color-developing substance, which was loaded on the diffusion path of a microporous strip, with Tween 20 as the moving substance. When the Tween 20 reached the test line of the PDA/SiO2 NC, the line transitioned from blue to red, indicating the TTI endpoint. A principal design factor was used to accurately locate the test line. Four TTIs (TTI5 degrees c, TTI10 degrees c,TTI15 degrees c, and TTI25 degrees c) were designed in combination with the various test line locations, by matching the TTI endpoints with the food deterioration time during storage at 5, 10, 15 and 25 degrees C, respectively. Wheatgrass juice, a healthy drink with a relatively short shelf-life at 5 degrees C of about 5 days, was used as the food specimen. During storage at 5-25 degrees C, TTI25 degrees c underpredicted the juice shelf life, such that the TTI reached the endpoint before the wheatgrass juice attained the deterioration point. However, the other TTIs did not always underpredict the juice shelf-life. The optimal diffusion-type TTI design was found to locate the test line of the PDA/SiO2 NC matching the shelf-life of the tested food at the highest possible temperature during storage. (C) 2017 Elsevier Ltd. All rights reserved.