The ovarian specimens obtained from the patients with 27 malignant germ cell tumors were analyzed in order to study DNA and form factor using flow cytometry and image analyzer. The malignant germ cell tumors consisted of six dysgerminomas, six endodermal sinus tumors, seven immature teratomas, three teratomas associated with squamous cell carcinoma, two embryonal carcinomas one mixed germ cell tumor and one malignant struma ovarii. Five normal ovaries used as control group. Various prognostic factors such as DNA ploidy, S-Phase and measurements of from were evaluated by flow cytometry and image analyzer. Mitotic index, histological grade, nuclear grade and tumor necrosis were assessed with microscope. These prognostic factors of DNA ploidy, S-phase, form factor histological were compared with tumor recurrence and clinical stage in this study. The results were as follows: 1. Sixteen(59%) out of 27 were aneuploidy and 11(41%) diploidy. All six dysgerminomas and six endodermal sinus tumors were aneuploidy, while six immature teratomas were diploidy. 2. S-phase fractions of malignant germ cell related significantly with those mitotic index(P=0.0201). S-phase fractions significantly increased in mitotic index gradeIII compared with grade I(P<0.01). There was no significant different between grade Iand II, and between gradeII and III. There was no difference between S-phase fractions and the remaining histological variables. 3. The incidence of aneuploidy was higher in the high s_ phase fractions(P=0.0041). However, there was no difference between S-phase fractions and tumor recurrence. 4. The incidence of aneuploidy significantly increased in clinical stage III and IV compared with stage I and II(P=0.0368). However, the difference between clinical stage histological variables was not significant. 5. The difference between form factor and histological variables, between from and tumor recurrence(P=0.3698), and between from factor and S-phase fractions(r=0.76) could not reach statistical significance. These results suggest that ploidy can give significant value for routine clinical prognostic prediction, whereas hitologic variables and form factor are poorly suitable for the prognostic evaluation.