The classification of extraovarian disease into invasive and noninvasive implants predicts patient outcome in patients with high-stage ovarian serous borderline tumors (tumors of low malignant potential). However, the morphologic criteria used to classify implants vary between studies. To date, there has been no large-scale study with follow-up data comparing the prognostic significance of competing criteria. Peritoneal and/or lymph node implants from 181 patients with high-stage serous borderline tumors were evaluated independently by 3 pathologists for the following 8 morphologic features: micropapillary architecture; glandular architecture; nests of epithelial cells with surrounding retraction artifact set in densely fibrotic stroma; low-power destructive tissue invasion; single eosinophilic epithelial cells within desmoplastic stroma; mitotic activity; nuclear pleomorphism; and nucleoli. Follow-up of 156 (86%) patients ranged from 11 to 264 months (mean, 89 mo; median, 94 mo). Implants with low-power destructive invasion into underlying tissue were the best predictor of adverse patient outcome with 69% overall and 59% disease-free survival (P<0.01). In the evaluation of individual morphologic features, the low-power destructive tissue invasion criterion also had excellent reproducibility between observers (κ=0.84). Extraovarian implants with micropapillary architecture or solid nests with clefts were often associated with tissue invasion but did not add significant prognostic value beyond destructive tissue invasion alone. Implants without attached normal tissue were not associated with adverse outcome and appear to be noninvasive. Because the presence of invasion in an extraovarian implant is associated with an overall survival analogous to that of low-grade serous carcinoma, the designation low-grade serous carcinoma is recommended. Even though the low-power destructive tissue invasion criterion has excellent interobserver reproducibility, it is further recommended that the presence of an invasive implant be confirmed by at least 2 pathologists (preferably at least 1 of whom is an experienced gynecologic pathologist) in order to establish the diagnosis of-low grade serous carcinoma.