Study design Retrospective cohort study. Objectives To compare Hounsfield unit (HU), vertebral bone quality (VBQ), and dual-energy X-ray absorptiometry (DEXA) T-score in predicting cage subsidence (CS) after posterior lumbar interbody fusion. Methods We retrospectively reviewed patients undergoing PLIF and measured L4 HU and VBQ. Baseline characteristics between CS and non-CS groups were compared. Multiple logistic regression analysis was used to identify risk factors for CS. Receiver operating characteristic (ROC) curves were used to explore the predictive value of variables for CS. Subgroup analysis was conducted for patients with available DEXA results. Results Eighty-four patients were analyzed. The subsidence group was older, with lower L4 HU (92.6 (35.4) vs 130.8 (47.8), P < 0.001) and higher VBQ (3.30 (0.64) vs 2.99 (0.63), P = 0.041) compared to the non-CS group. Low L4 HU (OR 0.979, P = 0.022) was an independent risk factor for subsidence, while VBQ was not. ROC analysis indicated that L4 HU (AUC = 0.733, P < 0.001) exhibited superior predictive performance compared to VBQ (AUC = 0.643, P = 0.032). Subgroup analysis in patients with DEXA showed that the femoral neck T-score (AUC = 0.67, P = 0.037) could differentiate CS, with L4 HU exhibiting the highest predictive ability (AUC = 0.778, P < 0.001), both outperforming VBQ (AUC = 0.645, P = 0.075). Conclusions Among the three BMD parameters, only low L4 HU could be an independent risk factor for CS. Preoperative HU is mandatory in patients at risk for osteoporosis when undergoing spine surgery.