The adding-on phenomenon is a common complication in adolescent idiopathic scoliosis (AIS) patients after correction surgery. However, the risk factors of previous studies and the optimal treatment strategies remain controversial. The aim of this study was to identify new risk factors for the adding-on phenomenon after posterior correction surgery in AIS patients and compare different treatment strategies to guide the selection of the lowest instrumented vertebra (LIV). All types of Lenke AIS patients who received correction surgery at our center from January 2009 to July 2014 were analyzed. The anteroposterior and lateral films were evaluated before surgery, at the 2-week follow-up, and at the 2-year or later follow-up. The patients were divided into 2 groups according to whether adding-on was observed at the last follow-up. The factors predictive of the adding-on phenomenon were identified in a multivariate binary logistic regression model. Different methods for LIV selection were compared in both the adding-on group and the control group (no adding-on). Out of the 346 patients reviewed, 92 met the inclusion criteria; 22 of these met the definition for distal adding-on, and were included in adding-on group. The remaining 70 patients were included in the no adding-on group. The average follow-up was 3.6 years. Touch classification (P<. 000), Dnfs (P=. 005), and vertebra number between LIV and angle velocity (AV) (P=. 001) were significantly different between the 2 groups. Age, gender, Risser sign, and screw density were not found to be affiliated with the presence of adding-on. The results of the Scoliosis Research Society (SRS)-22 were not significantly different between the adding-on group and the control group for any section or overall (P>. 05). Binary logistic regression results indicated that postoperative LAV deviation from the CSVL and Touch classification were independent predictive factors. Among the 4 methods, only choosing touch type A as LIV shows satisfactory outcome. The Touch classification is an important risk factor that is highly correlated with the incidence of the adding-on phenomenon. The best LIV choice to preserve the lumbar activity segment as much as possible is Touch type C, and no significant difference was observed in the SRS-22 scores between the Touch type C group and the control group in the short-term follow-up.