Histologic criteria for the distinction of ovarian mucinous borderline tumors (MBTs) from invasive mucinous carcinomas (MUCCAs) and the definitions of intraepithelial (noninvasive) carcinoma and microinvasion are controversial. Accurate assessment of the behavior of these tumors has been obscured by inclusion of cases of pseudomyxoma peritonei (PMP), an entity of extraovarian origin, and misclassification of the ovarian tumors in PMP and metastatic mucinous carcinomas (METCAs) as either advanced-stage MBTs or primary ovarian MUCCAs. One hundred thirty-six intestinal-type ovarian mucinous tumors without PMP were evaluated for the presence of stromal invasion, marked cytologic atypia, epithelial stratification of more than three cell layers, and necrosis. Criteria for the diagnosis of MBT, MBT with intraepithelial carcinoma, MBT with microinvasion (MIBT), MUCCA, and METCA were established and correlated with behavior. Twenty-three (59%) of 39 patients whose tumors had stromal invasion of more than 5 mm died of disease. Stromal invasion of more than 5 mm was the sole feature that correlated with a poor prognosis. In the absence of this feature, marked cytologic atypia, epithelial stratification of more than three layers, microinvasion (<5 mm), or necrosis did not have an adverse effect on survival. Tumors were classified as MBT (n = 65; 48%) based on absence of stromal invasion, regardless of degree of cytologic atypia or epithelial stratification; of these, 28 (43%) qualified as intraepithelial carcinoma based on epithelial stratification of more than three cell layers or marked cytologic atypia. Tumors with stromal invasion of less than 5 mm were classified as MIBT (n = 8; 6%). Tumors with stromal invasion of more than 5 mm were classified as MUCCA (n = 24; 18%). Tumors with a nodular pattern of stromal invasion, morphology inconsistent with ovarian origin, or a primary site elsewhere were classified as METCA (n = 35; 26%). Four tumors could not be definitively classified. Of the 86 cases with follow-up (median, 33 months) all MBTs (n = 44) and MIBTs (n = 6) were stage I, with 5-year survival rates of 100%. MUCCAs (n = 17) that were stage I had a 5-year survival rate of 91%; all patients with advanced-stage MUCCA died of disease. Five-year survival rate for METCAs (n = 19) was 11%. METCAs were more common than MUCCAs but can mimic MUCCAs and MBTs clinically and histologically. In the absence of a primary site elsewhere, METCA should be strongly suspected when there is bilateral surface involvement and a characteristic nodular pattern of invasion. It is important to recognize this pattern because 5-year survival rate for METCA (11%) was substantially less than that of MUCCA (all stages, 51%) and MBT (100%). Because all MBTs, regardless of degree of atypia or stratification, were stage I and benign, we prefer to designate them as atypical proliferative mucinous tumors. Marked cytologic atypia, epithelial stratification of more than three layers, and microinvasion (<5 mm) had no effect on the behavior of MBT. Noninvasive mucinous tumors with marked cytologic atypia or excessive epithelial stratification can be interpreted as atypical proliferative tumors with intraepithelial carcinoma and those with microinvasion can be designated as atypical proliferative tumors with microinvasion; these tumors appear to represent transitional stages in ovarian mucinous carcinogenesis.