Stage I, low-grade endometrioid and mucinous ovarian carcinomas have an excellent prognosis. Published data have suggested that destructive stromal invasion, a relatively uncommon finding in these tumors, is a poor prognostic factor. We investigated this by studying all FIGO stage I, grades 1 and 2 (of 3) endometrioid and mucinous ovarian carcinomas that were surgically staged at the Memorial Sloan-Kettering Cancer Center from 1980 to 2000. We undertook a careful review of all available slides using current diagnostic criteria and correlated histopathologic indices with clinical outcome data. Cases studied included 13 endometrioid ovarian carcinomas (stage IA, eight; stage IC, five) and six intestinal mucinous ovarian carcinomas (stage IA, three; stage IC, three). All of the tumors contained areas of expansile invasion, greater than that acceptable for microinvasion, and were thus diagnosed as carcinomas instead of borderline tumors. Nevertheless, nearly all demonstrated borderline tumor (noninvasive) components. Six tumors contained at least one focus of destructive stromal invasion (two endometrioid and four mucinous ovarian carcinomas). Four additional cases showed a focus suspicious for but not diagnostic of destructive invasion ('indeterminate for destructive invasion') (two endometrioid and two mucinous ovarian carcinomas). Follow-up data were available for 17 patients. The median follow-up was 81 months (range, 9-161 months). In all, 14 patients were alive with no evidence of disease (expansile invasion alone, eight; destructive stromal invasion, four; and indeterminate for destructive invasion, two). Three patients died of their disease (destructive stromal invasion, two; and indeterminate for destructive invasion, one). The size, number, and nuclear grade of destructive stromal invasion foci did not appear to have an impact on survival in this relatively limited number of patients. Outcome data in patients with stage I, low-grade endometrioid and mucinous ovarian carcinomas without destructive stromal invasion indicate that these tumors have a very limited malignant potential. The literature has not documented recurrences in this setting when the staging has been complete, the sampling adequate, and the cytologic features no more than grade 2, and metastasis to the ovary has been excluded. In contrast, carcinomas harboring even limited foci of destructive stromal invasion, although frequently cured surgically, can pursue a malignant clinical course.