The work aims to solve the problem in the quality control of QR code on cigarette packaging, to provide technical support for quality control of QR code. An orthogonal experiment was designed to determine the key production parameters affecting the quality level of QR code by variance analysis, and the quality of QR code was modeled by multiple linear regression algorithm. It was found that the symbol contrast, inherent graphic fouling and modulation index rating of QR code were the most unstable in production, which were the keys to quality control. Different substrate surface energy had an effect on droplet diffusion, and a quality prediction model of modulation and inherent pattern fouling was established. The importance of rating parameters and their affecting factors to quality control is emphasized, and the quality control of QR code is realized by modeling algorithm. It is pointed out that the quality prediction model is of great significance to support the quality control of QR code.