Based on the visible/nearinfrared local transmission spectrum, a handheld multi-quality non-destructive inspection device for potatoes was independently developed, including a spectrum acquisition module, a light source module, a control and display module, a power supply module, and a black and white reference box. The entire device size was 11cm×6.5cm×9.5cm. Totally eighty-five Atlantic potato samples were selected, and a partial leastsquares quantitative prediction model of potato water, starch and reducing sugar was established based on the developed device. The correlation coefficients of potato water, starch and reducing sugar prediction model validation set were 0.9278, 0.9146 and 0.9338, and the root mean square errors were 0.3253%, 0.3449% and 0.0416%, respectively. Based on the QT development tool, the device realtime analysis and control software was written in C/C++ language. The potato water, starch and reducing sugar prediction models were built into the software. The real-time nondestructive detection of potato multi-parameter parameters was realized by onebutton operation. Finally, the accuracy of the multi-quality nondestructive testing device for handheld potato was verified. The correlation coefficients between the predicted values of water content, starch and reducing sugar content of 20 potato samples and standard physical and chemical values were 0.9141, 0.9122 and 0.9140, the root mean square errors were 0.3527%, 0.3404% and 0.0400%, and the average deviations were 0.2951%, 0.2536% and 0.0316%, respectively. The results showed that the potato multi-quality local transmission non-destructive testing device can realize the real-time non-destructive detection of potato moisture.