We used the technique of image analysis to simultaneously measure DNA content and nuclear morphology of 21 borderline ovarian tumors. Aneuploidy was identified in 9 of 21 tumors and was unrelated to tumor stage or nuclear grade. Morphometric nuclear features that were measured included size, shape, texture, and average density. Nuclear size and shape were positively correlated (r = 0.507), and nuclear size and average density were negatively correlated (r = -0.772). Six tumors recurred and recurrence was significantly associated with tumor aneuploidy (P = 0.046), stage III tumors (P = 0.03), and increased nuclear texture (P = 0.07). These results suggest that measurement of DNA ploidy and nuclear morphology using image analysis can provide important prognostic information in patients with borderline ovarian tumors.