Objective. To evaluate how the independent predictors of recurrence f or stage IB2 cervical cancers treated with up-front radical hysterectomy apply to establ ished risk models. Methods. Patients with IB2 cervical cancers diagnosed from 19 90 to 2000 were identified from tumor registries of two institutions. Patients w ere classified into risk groups: high-risk (HR) (positive nodes, parametria, or margins), intermediate-risk (IR) (positive lymph vascular space involvement (L VSI) with any cervical stromal invasion (CSI), or (-) LVSI and > middle-CSI), or low-risk (LR) (absence of HR or IR characteristics). Disease free survival ( DFS) was estimated by Kaplan-Meier method and comparisons between subgroups wer e studied by log rank. A Cox proportional hazards model was used to determine in dependent predictors of recurrence. Results. We identified 86 patientswith stage IB2 tumors treated by RH.We found 34%of patients to be HR, 60%IR, and 6%LR. Of the 52 IR patients, 28 had (+) LVSI with superficial, middle, and outer 1/3 CSI, and 24 had (-) LVSI with middle or outer 1/3 invasion. Overall, postoperat ive adjuvant radiation (PRT) was used in 52%of the 86 patients, including 0/5 L R, 16/52 IR, and 29/29 HR patients. Univariate predictors of recurrence were pel vic nodal disease, (+) LVSI, (+) parametria, outer 1/3 CSI, and tumor size > 6 cm. Age, grade, histology, and the use of postoperative radiation were not asso ciated with recurrence. Multivariate analysis identified LVSI as the only indepe ndent predictor of recurrence (RR 5.2, P = 0.03). Two-year DFS for LR, IR, and HR patients was 100%, 83%, and 60%, respectively. Only 4/24 (17%) IR patient s with (-) LVSI got PRT compared with 12/28 (43%) of IR patients with (+) LVSI. The 2-year DFS for IR patients with (-) LVSI was 96%. IR (+) pa tients recurred more frequently with a 2-year DFS of 71%. Conclusions. Overall , 66%of patients with IB2 disease were classified as having low or intermediate -risk disease. IR patients with (-) LVSI and all LR patients did well with sur gery alone. This study defines the independent importance of LVSI and questions the utility of published IR models when applied to stage IB2 cervical cancer. Objective. To evaluate how the independent predictors of recurrence f or stage IB2 cervical cancers treated with up-front radical hysterectomy apply to establ ished risk models. Methods. Patients with IB2 cervical cancers diagnosed from 19 90 to 2000 were identified from tumor regstries of two Institutions. Patients w ere classified into risk groups: high-risk (HR) (positive nodes, parametria, or margins), intermediate-risk (IR) , or (-) LVSI and> middle-CSI), or low-risk (LR) (absence of HR or IR characteristics). Disease free survival (DFS) was estimated by Kaplan-Meier method and comparisons between subgroups w e e studied by Log rank. A Cox proportional hazards model was used to determine in dependent predictors of recurrence. Results Weighted 86 patients with stage IB2 tumors treated by RH.We found 34% of patients to be HR, 60% IR, and 6% LR. Of the 52 IR patients, 28 had (+) LVSI with Overall, postoperation ive adjuvant radiation (PRT) was used in 52% of the 86 patients, including 0/5 LR, 16/52 IR, and 29/29 HR patients. Univariate predictors of recurrence were pel vic nodal disease, (+) LVSI, (+) parametria, outer 1/3 CSI, and tumor size> 6 cm. , histology, and the use of postoperative radiation were not asso ciated with recurrence. Multivariate analysis identified LVSI as the only independent predictor of recurrence (RR 5.2, P = 0.03). Two-year DFS for LR, IR, and HR patients was Only 4/24 (17%) IR patient s with (-) LVSI got PRT compared with 12/28 (43%) of IR patients with (+) LVSI. The 2 (100%), 83%, and 60% -year DFS for IR patients with (-) LVSI was 96%. IR (+) pa tients recurred more frequently with a 2-year DFS of 71%. Conclusions. Overall, 66% of patients with IB2 disease were classified as having low or intermediate -risk disease. IR patients with (-) LVSI and all LR patients did well with sur gery alone. This study defines the independent importance of LVSI and questions the utility of published IR models when applied to stage IB2 cervical cancer.