Abstract: In order to improve sewage treatment efficiency, the data of 105 sewage treatment plants covering all prefecture-level cities in Shandong Province in 2018 were selected as samples. On the basis of data envelopment analysis (DEA), partial output indexes were compared as undesirable outputs by SBM-Undesirable models. This paper evaluated the efficiency of the plants and the key factors that contribute the low efficiency. The average sewage treatment efficiency was 0.64. The technical efficiency value and the scale efficiency value were 0.67 and 0.92, respectively, which are higher than the national average (the national average is 0.62, 0.62 and 0.96, respectively). Low technical efficiency was the main restriction on the comprehensive efficiency of the sewage treatment industry in Shandong Province. After setting undesirable output in the same input-output index, the overall efficiency is lower than that of the traditional DEA model, indicating that the discharge of COD, ammonia nitrogen and total phosphorus had a negative effect on sewage treatment. In the technical classification, the technical efficiency of the A²/O treatment process in large sewage treatment plants reached the highest value of 0.8. The technical efficiency of other activated sludge technologies in small sewage treatment plants was 0.7, and they had the advantage in treating COD and total phosphorus. The average technical efficiency of oxidation ditch treatment process was 0.55 and the total phosphorus treatment effect was poor, which is suitable for small and medium-sized sewage treatment enterprises. The technology types, scale and main pollutants have interactive effects on efficiency of sewage treatment. Shandong Province should improve the sewage treatment efficiency through the coordinated improvement of scale and technology according to the specific situation of each sewage treatment enterprise.