The work aims to ensure the constant tension of mark tapes used in the labeling machine to improve the labeling accuracy. A constant tension control method for labeling machine was put forward based on the single neuron adaptive algorithm. The dynamic model of label deliver mechanism of the labeling machine was analyzed in detail, and the dynamic torque balance equation was obtained. The key factors influencing constant tension were expounded. A tension adaptive control algorithm was proposed based on the single neuron control theory and the corresponding controller was designed. Finally, a test platform was set up to carry out the related experiment studies. The results showed that, based on the control method described herein, the labeling machine's standard-label rate was more than 97% and the tension was constant during the labeling. The proposed control system has advantages such as high labeling precision, wide application range and fast and stable response.