How much borderline change in an otherwise typical ovarian serous cystadenoma should warrant classification as a serous ovarian "borderline tumor?" We correlated estimated volume and percent borderline change with stage in 56 cases of serous ovarian neoplasms (excluding carcinomas) diagnosed as at least focal borderline change to see if we could define an appropriate threshold for the diagnosis of borderline tumor that would justify full surgical staging. Forty-three cases were completely staged, 6 had "fertility-sparing" but otherwise complete staging, and 7 cases had "limited" staging. Thirty-eight cases were stage 1a-1c, and 18 were greater than stage 1. Cases with stage 1 disease had a significantly lower mean volume of borderline change sampled of 2.0 compared with 5.6 cm in cases with greater than stage 1 disease (P = 0.0002). All high-stage cases had at least 1.0 cm or more of borderline change sampled (range, 1.0-12). Cases with stage 1 disease had a significantly lower mean estimated total percent borderline change of 34.8% compared with 77.2% in cases with greater than stage 1 disease (P < 0.0001). All high-stage cases had 20% or more total borderline change (range, 20%-100%). In addition, a grossly exophytic growth pattern component was highly predictive of high stage (P < or = 0.0001). Two cases recurred-both were advanced-stage and high-percent borderline change. There were no deaths due to disease (mean follow-up, 85 months). Our study supports a conservative 10% cutoff for classification as a "borderline tumor," and that complete surgical staging is not necessary when a serous neoplasm with an intracystic growth pattern has less than 10% or 0.5-cm borderline change.