To investigate the factors associated with proximal junctional kyphosis (PJK). A systematic search was performed. The weighted mean difference (WMD) was pooled for continuous variables, and the odds ratio (OR) was calculated for dichotomous variables. The PJK group had higher values for age (WMD = 2.53, 95%CI = 1.38 ~ 3.68, P < 0.001), female gender (OR = 1.56, 95%CI = 1.29 ~ 1.87, P < 0.001), and diagnosed osteoporosis (OR = 1.58, 95%CI = 1.11 ~ 2.26, P = 0.01). Preoperatively, significant differences were detected in sagittal vertical axis (SVA) (WMD = 19.29, 95%CI = 16.60 ~ 21.98, P < 0.001), pelvic incidence minus lumbar lordosis (PI-LL) (WMD = 2.71, 95%CI = 0.25 ~ 5.18, P = 0.03), pelvic tilt (PT) (WMD = 2.64, 95%CI = 1.38 ~ 3.90, P < 0.001), lumbar lordosis (LL) (WMD = - 1.76, 95%CI = - 2.73 ~ -0.79, P < 0.001), and sacral slope (SS) (WMD = - 2.80, 95%CI = - 5.57 ~ -0.04, P = 0.001). At follow-up, the following were higher in the PJK group: thoracic kyphosis (TK) (WMD = 5.51, 95%CI = 2.23 ~ 8.80, P < 0.001), proximal junctional angle (PJA) (WMD = 9.07, 95%CI = 4.21 ~ 13.92, P < 0.001), and PT (WMD = 1.51, 95%CI = 0.31 ~ 2.72, P = 0.01). However, there was no significant difference in SS (P = 0.49), and SVA (P = 0.11) between groups. Fusion to S1 or pelvis significantly increased the risk of PJK (OR = 2.08, P < 0.001). Ligament augmentation reduced the risk of PJK (OR = 0.34, 95%CI = 0.21 ~ 0.53, P < 0.001) better than the use of laminar hook (OR = 0.46, P < 0.001). Although no difference was detected for preoperative SRS-22 score (P = 0.056), a lower score (WMD = - 0.24, 95%CI = - 0.35 ~ -0.14, P < 0.001) was detected in PJK group at follow-up. The elderly female ASD patients were more susceptible to PJK, especially for those with osteoporosis, high preoperative SVA, low LL, large PT, and LIV extended to pelvis. The use of laminar hook and ligament reinforcement at the proximal end might prevent PJK.