Growing insights into the biological features and molecular underpinnings of ovarian cancer has prompted a shift toward histotype-specific treatments and clinical trials. As a result, the preoperative diagnosis of ovarian carcinomas based on small tissue sampling is rapidly gaining importance. The data on the accuracy of ovarian carcinoma histotype-specific diagnosis based on small tissue samples, however, remains very limited in the literature. Herein, we describe a prospective series of 30 ovarian tumors diagnosed using cytology, frozen section, core needle biopsy, and immunohistochemistry (p53, p16, WT1, HNF-1β, ARID1A, TFF3, vimentin, and PR). The accuracy of histotype diagnosis using each of these modalities was 52%, 81%, 85%, and 84% respectively, using the final pathology report as the reference standard. The accuracy of histotype diagnosis using the Calculator for Ovarian Subtype Prediction (COSP), which evaluates immunohistochemical stains independent of histopathologic features, was 85%. Diagnostic accuracy varied across histotype and was lowest for endometrioid carcinoma across all diagnostic modalities (54%). High-grade serous carcinomas were the most overdiagnosed on core needle biopsy (accounting for 45% of misdiagnoses) and clear cell carcinomas the most overdiagnosed on frozen section (accounting for 36% of misdiagnoses). On core needle biopsy, 2/30 (7%) cases had a higher grade lesion missed due to sampling limitations. In this study, we identify several challenges in the diagnosis of ovarian tumors based on limited tissue sampling. Recognition of these scenarios can help improve diagnostic accuracy as we move forward with histotype-specific therapeutic strategies.