Objective : Bone cement leakage is a well-known potential complication of percutaneous vertebroplasty (PVP) in patients with osteoporotic compression fracture. Even though there has been a controversy in the efficacy of antecedent venography to prevent this complication, many authors have performed intra osseous venography before bone cement injection. The goal of this study was to classify the venous drainage patterns of spine before PVP, and compare their patterns at different vertebral levels. Methods : The authors retrospectively reviewed 1,042 intraosseous venographic patterns in 321 patients with 574 osteoporotic compression fractures during six-year period in one institution. To classify venogram patterns, we selected simple lateral X-ray of spine taken immediately after injection of the contrast dye. We classified the venography patterns according to contrast leakage pattern and leakage direction as follows; trabecular (TR), trabecular anterior (TA), trabecular posterior (TP), trabecular anterior-posterior (TAP), trabecular lateral (TL), venous anterior(VA), venous posterior (VP), venous anterior-posterior (VAP), soft tissue (ST). Also, we compared venogram patterns according to different spinal levels. Results : In overall, the most common pattern was TP type accounting for 37.4% (390/1042) of all intraosseous venograms. This is followed by TAP in 21.5%, TR 17.4%, TA 116%, TL 5.8%, ST 4.1%, VA 1.2%, VP 0.6%, and VAP 0.4% in descending order of frequency. According to the spinal level, TR and TAP types were most common in thoracic spine (T6-T10), TP type was most common in thoraco-Iumbar spine (T11-L2), and TP and TAP types were most common in lumbo-sacral spine (L3-S1). Contrast dye leakage to soft tissue such as psoas muscle or disc were detected in 43 (4.1%) venograms. Direct venous drainage without staining of vertebral body was found in 23 (2.2%) venograms. The 8.3% of thoracic venogram showed direct venous drainage. Thoracic level showed a more tendency of direct venous drainage than other spine levels (p<001). Conclusion : The authors propose a new classification system of intra osseous venography during PVP. The trabecular-posterior (TP) type is most common through all spine, and venous-filling (V) type was most frequent in thoracic spine. Further study would be necessary to elucidate the efficacy of this classification system to prevent bone cement leakage during PVP.