The role of endogenous androgens and sex hormone-binding globulin (SHBG) in ovarian carcinogenesis is poorly understood. Epithelial invasive ovarian cancer (EOC) is a heterogeneous disease and there are no prospective data on endogenous androgens and EOC risk by tumor characteristics (histology, grade, stage) or the dualistic model of ovarian carcinogenesis (i.e. type I vs. type II, leading to less or more aggressive tumors). We conducted a nested case-control study in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort evaluating androgens and SHBG and invasive EOC risk by tumor characteristics. Female participants who provided a blood sample and were not using exogenous hormones at blood donation were eligible (n = 183,257). A total of 565 eligible women developed EOC; two controls (n = 1,097) were matched per case. We used multivariable conditional logistic regression models. We observed no association between androgens, SHBG and EOC overall. A doubling of androstenedione reduced risk of serous carcinomas by 21% (odds ratio (OR)log2=0.79, 95% confidence interval [CI]=[0.64-0.97]). Moreover, associations differed for low-grade and high-grade carcinomas, with positive associations for low-grade and inverse associations for high-grade carcinomas (e.g. androstenedione: low grade: ORlog2=1.99 [0.98-4.06]; high grade: ORlog2=0.75 [0.61-0.93], p(het)0.01), similar associations were observed for type I/II tumors. This is the first prospective study to evaluate androgens, SHBG and EOC risk by tumor characteristics and type I/II status. Our findings support a possible role of androgens in ovarian carcinogenesis. Additional studies exploring this association are needed. What's new? There appear to be several types of epithelial invasive ovarian cancer (EOC), and hormone-related risk factors are poorly understood. In this study, the authors found that the impact of endogenous androgens on the risk of developing EOC differed depending upon tumor characteristics. Androgen concentrations were positively associated with the risk of low-grade and type-I carcinomas, but the study found an inverse association for high-grade tumors. These findings support a possible role for androgens in ovarian carcinogenesis, and emphasize the need for additional research.