Abstract: At present, more and more pipelines adopt the operation mode of adding drag reduction agent (DRA). A good prediction model of drag reduction rate can provide the guidance for on-site operation of adding DRA and avoid unnecessary waste. In this paper, a prediction model of drag reduction rate was established based on the method in the mixing length theory for solving the turbulence additional stress and the single-factor empirical equation of the negative index of drag reduction rate, after the relationships between the friction coefficient and turbulence additional stress before and after the DRA addition were analyzed. Then, for FLO MXA of Baker Hughes, Liquid Power TM of ConocoPhillips, ks-30-01 and JIA Species DRA, the decline rate and flow characteristic coefficient in the model were determined using a large number of on-site operation data. In addition, the same DRA was applied to different pipelines for pipe diameter and viscosity correction, and accordingly the correction coefficient was obtained. It is indicated that the proposed prediction model can be used for the prediction of different pipes and different DRAs, and the prediction accuracy is high. The relative error of the prediction results remains within 10% and the minimum valve can be as low as zero.