The work aims to predict the surface roughness and thickness of gravure printing products to improve the quality and speed of printing patterns, in order to realize the rapid intelligent printing. The mathematical model including such independent variables as operation tension, printing speed, ink viscosity and theoretical transfer volume was established by means of 24 full factorial experiment design method. Pareto statistical analysis was applied to analyze the interaction between the variables. The results showed that the surface roughness of printing pattern was high, with the maximum and minimum values of 1.812 μm and 1.524 μm, respectively. The determination coefficients of the mathematical models for thickness and surface roughness of gravure printing were 76.32% and 82.25%, respectively. By considering other parameters such as scraper type and pressure zone pressure, the determination coefficient (R2) of the model could be improved, thus improving the model accuracy. Under the operating conditions of high tension (4.0 N), the surface roughness of the theoretical transfer volume (1.2 cm3/m2) was more uniform, and the use of low viscosity ink would lead to a greater image thickness and then reduce the manufacturing cost. The major influencing factors for thickness change are viscosity, speed, tension*speed, speed*viscosity and tension* viscosity*theoretical transfer volume (* indicates the combined action of several factors). The major influencing factors for surface roughness are theoretical transfer volume, tension*theoretical transfer volume, tension*speed, tension, viscosity*theoretical transfer volume. The most significant factor in determining the surface roughness is the theoretical transfer volume.